Uploaded by Behrouz Rezaei

ar10-6

advertisement
DOT/FAA/AR-10/6
Air Traffic Organization
NextGen & Operations Planning
Office of Research and
Technology Development
Washington, DC 20591
Determining the Fatigue Life of
Composite Aircraft Structures
Using Life and Load-Enhancement
Factors
June 2011
Final Report
This document is available to the U.S. public
through the National Technical Information
Services (NTIS), Springfield, Virginia 22161.
This document is also available from the
Federal Aviation Administration William J. Hughes
Technical Center at actlibrary.tc.faa.gov.
U.S. Department of Transportation
Federal Aviation Administration
NOTICE
This document is disseminated under the sponsorship of the U.S.
Department of Transportation in the interest of information exchange. The
United States Government assumes no liability for the contents or use
thereof. The United States Government does not endorse products or
manufacturers. Trade or manufacturer's names appear herein solely
because they are considered essential to the objective of this report. The
findings and conclusions in this report are those of the author(s) and do
not necessarily represent the views of the funding agency. This document
does not constitute FAA policy. Consult the FAA sponsoring organization
listed on the Technical Documentation page as to its use.
This report is available at the Federal Aviation Administration William J.
Hughes Technical Center’s Full-Text Technical Reports page:
actlibrary.act.faa.gov in Adobe Acrobat portable document format (PDF).
Technical Report Documentation Page
1. Report No.
2. Government Accession No.
3. Recipient's Catalog No.
DOT/FAA/AR-10/6
4. Title and Subtitle
5. Report Date
DETERMINING THE FATIGUE LIFE OF COMPOSITE AIRCRAFT
STRUCTURES USING LIFE AND LOAD-ENHANCEMENT FACTORS
June 2011
7. Author(s)
8. Performing Organization Report No.
6. Performing Organization Code
John Tomblin and Waruna Seneviratne
9. Performing Organization Name and Address
10. Work Unit No. (TRAIS)
Department of Aerospace Engineering
National Institute for Aviation Research
Wichita State University
Wichita, KS 67260-0093
11. Contract or Grant No.
12. Sponsoring Agency Name and Address
13. Type of Report and Period Covered
U.S. Department of Transportation
Federal Aviation Administration
Air Traffic Organization NextGen & Operations Planning
Office of Research and Technology Development
Washington, DC 20591
Final Report
14. Sponsoring Agency Code
ACE-110
15. Supplementary Notes
The Federal Aviation Administration William J. Hughes Technical Center COTRs were Curtis Davies and Peter Shyprykevich.
16. Abstract
Current regulations require airframes to demonstrate fatigue service life by testing and/or analysis. This requirement is identical
for both metals and composites. As composites exhibit higher scatter than metals, extensive scatter analysis was conducted by
Northrup/Grumman using several material databases. These data represented several structural details and variables such as
laminate lay-up, loading mode, load transfer, specimen geometry, and environment. The U.S. Navy program included only
autoclaved 350°F-cure graphite-epoxy materials, and the analysis was conducted primarily on fiber-dominated failures of
laminated construction and did not include sandwich construction or bonded joints. The data analysis of this program developed
a load-enhancement factor (LEF) that shortened the fatigue testing time while maintaining equivalent reliability to that of metals.
The research performed in this study extended the developed methodology to new material systems and construction techniques.
This study included data for materials commonly used in aircraft applications, including adhesives and sandwich construction.
Testing consisted of various element-type tests and concentrated on tests that were generic in nature and were representative of
various loading modes and construction techniques. In addition, the database available at the National Institute of Aviation
Research (NIAR) was included to expand the data for the scatter analysis. Three different techniques are discussed for scatter
analysis of fatigue data: individual Weibull, joint Weibull, and the Sendeckyj wearout model. It is recommended that the analysis
include specimens or elements representative of features of a particular structure, i.e., materials, design details, failure modes,
loading conditions, environments, rather than pool various material databases. Also, it is noted that the primary goal in scatter
analysis is not to select shape parameters from the critical lay-up, R-ratio, environment, etc. (which may result in skewed data that
will produce unconservative LEF), but rather to select the design details representing the critical areas of the structure. When
designing test matrices for generating fatigue shape parameters, it is essential to investigate the design details and conditions that
produce the most data scatter, which consequently affects the reliability of test data in higher levels of building blocks of testing.
The examples in this report show that the LEF are lower for modern composite structures. However, care must be taken to ensure
the integrity of test matrices to produce safe and reliable certification requirements. This report provides guidance on generating
shape parameters for calculating life factor and LEFs as well as application of these factors to a fatigue load spectrum without
compromising or significantly altering the fatigue life of the structure, especially in the case of hybrid (composite and metal)
structures.
17. Key Words
18. Distribution Statement
Scatter analysis, Life factor, Load-enhancement factor, Test
data, Composite fatigue, Weibull statistics
This document is available to the U.S. public through the
National Technical Information Service (NTIS), Springfield,
Virginia 22161. This document is also available from the
Federal Aviation Administration William J. Hughes Technical
Center at actlibrary.tc.faa.gov.
19. Security Classif. (of this report)
Unclassified
Form DOT F1700.7 (8-72)
20. Security Classif. (of this page)
Unclassified
Reproduction of completed page authorized
21. No. of Pages
155
22. Price
ACKNOWLEDGEMENT
The authors would like to acknowledge the technical guidance and support of Mr. Curtis Davis,
Dr. Larry Ilcewicz, and Mr. Peter Shyprykevich (retired) of the Federal Aviation Administration.
The support of the Composites and Structures Laboratories of the National Institute for Aviation
Research, especially test support from Govind Ramakrishna Pillai and Jason Yeoh and program
support from Upul Palliyaguru, is greatly appreciated. The authors would also like to thank
Cessna Aircraft of Wichita, Kansas, and Toray Composites (America), Inc., Tacoma,
Washington, for providing material support during specimen fabrication, and Liberty Aerospace
of Melbourne, Florida, for sharing their test data.
iii/iv
TABLE OF CONTENTS
Page
xv
EXECUTIVE SUMMARY
1.
2.
INTRODUCTION
1
1.1
1.2
6
8
EXPERIMENTAL PROCEDURE
2.1
2.2
2.3
3.
9
Material Systems
Test Matrices
Experimental Setup
10
10
13
2.3.1
2.3.2
2.3.3
2.3.4
2.3.5
13
15
15
17
17
Impact Tests
Nondestructive Inspections
Full-Field Strain Evolution
Static and Residual Strength Tests
Fatigue Life Evaluation
ANALYSIS OF COMPOSITE TEST DATA
18
3.1
Scatter Analysis
19
3.1.1 Individual Weibull Method
3.1.2 Joint Weibull Method
3.1.3 Sendeckyj Equivalent Static-Strength Model
20
21
21
Life-Factor Approach
Load-Factor Approach
Combined Load-Life Approach
22
25
28
3.4.1 Application of LEF to a Load Spectrum
3.4.2 Generating Life Factor and LEFs
31
35
Scatter Analysis Computer Code
37
3.2
3.3
3.4
3.5
4.
Background for Current Approach
Research Objectives and Overview
STATIC STRENGTH DATA SCATTER ANALYSIS
38
4.1
Structural Details for Static Scatter Analysis
39
4.1.1
Advanced Composites Group AS4/E7K8 3K Plain-Weave Fabric
40
4.1.2
Toray T700/#2510 Plain-Weave Fabric
44
v
4.2
5.
4.1.3
Toray 7781/#2510 8-Harness Satin-Weave Fabric
48
4.1.4
Toray T700/#2510 Unidirectional Tape
50
4.1.5
Advanced Composites Group AS4C/MTM45 Unidirectional Tape
51
4.1.6
Advanced Composites Group AS4C/MTM45 5-Harness
Satin-Weave Fabric
52
4.1.7
Nelcote T700/E765 Graphite Unidirectional Tape
54
4.1.8
Nelcote T300/E765 3K Plain-Weave Fabric
55
4.1.9
Adhesive Effects of Defects Data
56
Summary
57
FATIGUE LIFE DATA SCATTER ANALYSIS
59
5.1
5.2
Structural Details for Fatigue Life Scatter Analysis
Fatigue Scatter Analysis
59
60
5.2.1 The AS4/E7K8 Plain-Weave Fabric
5.2.2 The T700/#2510 Plain-Weave Fabric
5.2.3 The 7781/#2510 8-Harness Satin-Weave Fabric
5.2.4 Adhesive Fatigue Data
61
69
71
74
Summary of Fatigue Scatter Analysis
Load-Enhancement Factor
Case Study—Liberty Aerospace XL2 Fuselage
Case Study—Scatter Analysis of Bonded Joints
76
78
82
88
5.3
5.4
5.5
5.6
6.
CONCLUSIONS AND RECOMMENDATIONS
91
7.
REFERENCES
93
APPENDICES
A—Fatigue Test Results of Federal Aviation Administration Load-Enhancement Factor
Database
B—Scatter Analysis Results for Federal Aviation Administration Load-Enhancement
Factor Database
C—Scatter Analysis for Life and Load-Enhancement Factors
vi
LIST OF FIGURES
Figure
Page
1
Composite Materials Applications in Commercial Aircraft
1
2
Material Distribution for U.S. Navy F-18 Aircraft
3
3
Overview of Research
9
4
Instron Dynatup Drop-Weight Tester
14
5
Support Fixture for CAI and TAI Impact Specimens
14
6
The TTU Scanning of PFS Test Specimen
15
7
Portable Version of ARAMIS Photogrammetry System
16
8
Damage Evolution of a CAI Specimen Under Static Loading
16
9
The MTS Servohydraulic Test Frame
17
10
Compliance Change and Damage Area During Fatigue Tests of 4PB
Sandwich Specimen
18
11
Life Scatter in Composites and Metal
19
12
Scatter Analysis Using Weibull Distribution of Shape Parameters
20
13
Life-Factor Approach for Substantiating B-Basis Design Life
23
14
Influence of Fatigue Shape Parameter on B-Basis Life Factor
24
15
Influence of Strength and Life Parameter on LEF
28
16
Influence of MSSP and MLSP on LEF (Combined Load-Life Approach)
29
17
Combined Load-Life Approach for Composite Structures
30
18
A- and B-Basis LEF Requirements With Respect to the Number of Test Specimens
31
19
Application of Combined Load-Life Approach
32
20
Fatigue Test Spectrum Development for Composite Structural Test
33
21
Application of LEF Only to Mean Load
35
22
Minimum Test Requirements for Generating Life Factors and LEFs
36
vii
23
Minimum Requirements to be Fulfilled Prior to Using LEFs in Figure 18 for a
Composite Structural Test
37
24
Scatter Analysis Using SACC
38
25
Probability Density Function and Reliability Plot of Shape Parameters
for AS4-PW Static Strength Distributions
43
26
Shape Parameters for T700-PW Static-Strength Distributions
47
27
Shape Parameters for T700-UT Static-Strength Distributions
51
28
Shape Parameters for AS4C-UT Static-Strength Distributions
52
29
Shape Parameters for AS4C-5HS Static-Strength Distributions
53
30
Shape Parameters for E765-UT Static-Strength Distributions
54
31
Shape Parameters for E765-PW Static-Strength Distributions
55
32
Comparison of Composite Strength Shape Parameters for Different Environments
57
33
Comparison of Composite Strength Shape Parameters for Different Lay-Ups
58
34
Comparison of Composite Strength Shape Parameters for Different Loading Modes
58
35
Sendeckyj Wearout Analysis Prediction of Fatigue Life of OHT (R = 0) – AS4-PW
61
36
Effects of Lay-Up Sequence, AS4/E7K8, OH Measured Fatigue Data
63
37
Effects of Lay-Up Sequence, AS4/E7K8, OH Normalized Fatigue Data
63
38
Effects of Stress Ratio for AS4-PW OH Measured Fatigue Data
64
39
Effects of Stress Ratio for AS4-PW OH Normalized Fatigue Data
64
40
Effects of Lay-Up Sequence for AS4-PW CAI Normalized Fatigue Data
65
41
Comparison of CAI and TAI for AS4-PW
65
42
Fatigue-Life Shape Parameters of AS4-PW From Different Analysis Methods
66
43
Comparison of Static Strength and CV of AS4-PW From Test and
Sendeckyj Analysis
68
44
Effects of Stress Ratio for T700-PW OH Measured Fatigue Data
69
45
Effects of Stress Ratio for T700-PW OH Normalized Fatigue Data
70
viii
46
Fatigue-Life Shape Parameters of T700-PW From Different Analysis Methods
71
47
Effects of Stress Ratio for 7781-8HS OH Measured Fatigue Data
72
48
Effects of Stress Ratio for 7781-8HS OH Normalized Fatigue Data
72
49
Fatigue-Life Shape Parameters of 7781-8HS From Different Analysis Methods
73
50
Fatigue-Life Shape Parameters of FAA-D5656 Data From Different
Analysis Methods
75
51
Comparison of Fatigue-Life Shape Parameter for FAA-LEF Database
77
52
Influence of Test Duration on B-Basis LEFs for Different Materials
80
53
Influence of Test Duration on B-Basis LEFs of AS4-PW From Different
Analytical Techniques
81
54
Liberty Aerospace XL2 Aircraft
82
55
Influence of Test Duration on B-Basis LEFs for Liberty XL2
85
56
Comparison of Tested and Required LEFs for Liberty XL2
87
57
Composite-Titanium Bonded Joint Test Specimen
88
58
Comparison of LEFs Generated Using Composite and Adhesive Test Data
91
ix
LIST OF TABLES
Page
Table
1
Laminate Configurations for FAA-LEF Database
11
2
The FAA-LEF Test Methods and Fixture Requirements
11
3
The Basic FAA-LEF Test Matrix
12
4
Supplemental FAA-LEF Test Matrix
13
5
A-Basis LEFs
26
6
B-Basis LEFs
27
7
Static-Strength Test Results for AS4-PW
40
8
Weibull Parameters for Static-Strength Distributions of AS4-PW
41
9
Weibull Statistics for Combined Distribution of Scatter in Static-Strength
Distributions of AS4-PW
42
10
Static-Strength Test Results for T700-PW
44
11
Weibull Parameters for Static-Strength Distributions of T700-PW
44
12
Weibull Parameters for Static-Strength Distributions of
40/20/40 T700-PW (FAA-LVM)
45
Weibull Parameters for Static-Strength Distributions of
25/50/25 T700-PW (FAA-LVM)
46
Weibull Parameters for Static-Strength Distributions of
10/80/10 T700-PW (FAA-LVM)
47
15
Summary of Weibull Shape Parameter Analysis of T700-PW
48
16
Static-Strength Test Results for 7781-8HS
48
17
Weibull Parameters for Static-Strength Distributions of 7781-8HS
49
18
Weibull Statistics for Combined Distribution of 7781-8HS
49
19
Summary of Weibull Shape Parameter Analysis of T700-UT
50
20
Summary of Weibull Shape Parameter Analysis of AS4C-UT
51
21
Summary of Weibull Shape Parameter Analysis of AS4C-5HS
53
13
14
x
22
Summary of Weibull Shape Parameter Analysis of E765-UT
54
23
Summary of Weibull Shape Parameter Analysis of E765-PW
55
24
Weibull Parameters for Bonded-Joint PFS Element Tests
56
25
Weibull Parameters for Bonded SLS Element Tests
56
26
Fatigue-Life Scatter Analysis for AS4-PW
62
27
Comparison of Static Strength of AS4-PW From Test and Sendeckyj Analysis
67
28
Weibull Statistics for Combined Distribution of AS4-PW
68
29
Fatigue-Life Scatter Analysis for T700-PW
70
30
Weibull Statistics for Combined Distribution of T700-PW
71
31
Fatigue-Life Scatter Analysis for 7781-8HS
73
32
Weibull Statistics for Combined Distribution of 7781-8HS
74
33
Fatigue-Life Scatter Analysis for FAA-D5656
74
34
Weibull Statistics for Combined Distribution of FAA-D5656 Data
76
35
Weibull Statistics for Pooled Composite and Adhesive Data From Different
Analytical Techniques
78
36
Weibull Statistics for Combined Composites and Adhesives
79
37
Weibull Statistics for AS4-PW Composites and Adhesives From Different
Analytical Techniques
81
38
Weibull Analysis Results of Liberty Sandwich Static Test Data
82
39
Sendeckyj Analysis Results of Liberty Sandwich Fatigue Test Data
83
40
Comparison of Weibull Statistics for Liberty XL2 Database
84
41
Weibull Parameter Estimates for Static Tests
89
42
Weibull Parameter Estimates for Fatigue Tests
89
43
Summary of Strength and Life Shape Parameters of Composite and Adhesive Data
90
xi
LIST OF ACRONYMS
4PB
ACG
BVID
CAI
CFPW
CTD
CTU
CV
DaDT
DLL
DLT
DNC
EOD
ETD
ETW
FAA
FGSW
FH
LEF
LID
LLD
LVM
MIA
MLE
MLSP
MSSP
MTS
NASA
NDI
NIAR
OH
OHC
OHT
PFS
RRX
RRY
RTA
RTW
SACC
SLS
SLS-C
SLS-T
SMT
S/N
Four-point bend
Advanced Composites Group
Barely visible impact damage
Compression after impact
Carbon fabric plain weave
Cold temperature dry
Carbon tape unidirectional
Coefficient of variation
Durability and damage tolerance
Design limit load
Design lifetime
Double-notched compression
Effects of defects
Elevated temperature dry
Elevated temperature wet
Federal Aviation Administration
E-glass satin weave
Filled hole
Load-enhancement factor
Large impact damage
Load-life damage
Lamina variability method
Mechanical impedance analysis
Maximum likelihood estimation
Modal fatigue-life shape parameter
Modal static-strength shape parameter
Material Test Systems
National Aeronautics and Space Administration
Nondestructive inspection
National Institute for Aviation Research
Open hole
Open-hole compression
Open-hole tension
Picture-frame shear
Rank regression in X
Rank regression in Y
Room temperature ambient
Room temperature wet
Scatter Analysis Computer Code
Single-lap shear
Single-lap shear compression
Single-lap shear tension
Shear-moment torque
Stress to number of cycles
xii
TAI
TTU
UNI
VID
VNRS
Tension after impact
Through-transmission ultrasonic
Unidirectional lay-up
Visible impact damage
V-notch rail shear
xiii/xiv
EXECUTIVE SUMMARY
Over the past 25 years, the use of advanced composite materials in aircraft primary structures has
increased significantly. In 1994, with the Advanced General Aviation Transport Experiments
program, the National Aeronautics and Space Administration and the Federal Aviation
Administration revitalized the use of composites in general and commercial aviation. Driven by
the demand for fuel-efficient, light-weight, and high-stiffness structures that have fatigue
durability and corrosion resistance, modern large commercial aircraft are designed with more
than 50 percent composite materials. Although there are key differences between metal and
composite damage mechanics and durability concerns, the certification philosophy for
composites must meet the same structural integrity, safety, and durability requirements as that of
metals. Despite the many advantages, composite structural certification becomes challenging
due to the lack of experience in large-scale structures, complex interactive failure mechanisms,
sensitivity to temperature and moisture, and scatter in the data, especially in fatigue. The overall
objective of this research was to provide guidance into structural substantiation of composite
airframe structures under repeated loads through an efficient approach that weighs both the
economic aspects of certification and the timeframe required for testing, while ensuring safety.
The research methodology reported here consisted of combining existing certification
approaches used by various aircraft manufacturers with protocols for applying these
methodologies. This will permit extension of the methodologies to new material systems and
construction techniques.
This study included data for materials commonly used in aircraft applications, including
adhesives and sandwich construction. Testing consisted of various element-type tests and
concentrated on tests that were generic in nature and were representative of various loading
modes and construction techniques. In addition, the database available at the National Institute
of Aviation Research was included to expand the data for the scatter analysis. Three different
techniques were used for scatter analysis of fatigue data: individual Weibull, joint Weibull, and
the Sendeckyj wearout model. Procedures to generate reliable and economical scatter and loadenhancement factors necessary for a particular structural test by selecting the design details
representing the critical areas of the structure is outlined with several examples and case studies.
The effects of laminate stacking sequence, test environment, stress ratios, and several design
features, such as sandwich and bonded joints on the static-strength and fatigue-life shape
parameters, are discussed with detailed examples. Furthermore, several analytical techniques for
obtaining these shape parameters are discussed with examples. Finally, the application of loadenhancement factors and life factors for a full-scale test spectrum without adversely affecting the
fatigue life and the damage mechanism of the composite structure is discussed.
xv/xvi
1. INTRODUCTION.
In the early 1970s, composite materials were introduced to airframe structures to increase the
performance and life of the airframe. In 1977, the National Aeronautics and Space
Administration (NASA) Advanced Composite Structures Program introduced the use of
composites in primary structures in commercial aircraft, i.e., the Boeing 737 horizontal stabilizer
[1]. In 1994, the Advanced General Aviation Transport Experiments consortium, led by NASA
and supported by the Federal Aviation Administration (FAA), industry, and academia,
revitalized composite material product development in general aviation by developing costeffective composite airframe structures. Modern improved composite materials and matured
processes have encouraged commercial aircraft companies to increase the use of composites in
primary and secondary structures. Driven by the demand for fuel-efficient, light-weight, and
high-stiffness structures that have fatigue durability and corrosion resistance, the Boeing 787
Dreamliner is designed with more than 50 percent composite structure, marking a striking
milestone in composite usage in commercial aviation. Meanwhile, the Airbus A350 commercial
airplane is being designed with a similar percentage of composite materials in its structure.
Figure 1 shows the use of composites in several commercial aircraft applications.
Figure 1. Composite Materials Applications in Commercial Aircraft
1
Although there are key differences between metal and composite damage mechanics and
durability concerns, the certification philosophy for composites must meet the same structural
integrity, safety, and durability requirements as for metal aircraft. Over the years, composite and
hybrid structural certification programs have adopted methodologies used for metal structures
that are based on several decades of experience in full-scale structural certification and service.
Despite the advantages, such as high specific weight, tailorability, and fatigue resistance,
composite structural certification becomes challenging due to the lack of experience with largescale structures, complex interactive failure mechanisms, sensitivity to temperature and moisture,
and scatter in the data, especially relative to fatigue.
Most current fatigue life assessment methodologies for advanced composite structures rely on
empirical stress to number of cycles (S/N) data in the lower levels of the building block.
Variation of material characteristics for different fiber-resin systems, lay-up configurations,
environments, loading conditions, etc., often make the analysis and testing of composites
challenging. Anisotropic heterogeneous characteristics and a change in failure modes over the
fatigue life, as well as multiple failure mechanisms that interact with each other, make it
challenging to predict damage growth in composite structures. Consequently, most of the
damage mechanisms and wearout approaches (discussed later in this section) also depend on
empirical data for refinement or calibration. Some approaches only discuss failure progression
under certain loading configurations and often specific to a material system. Fatigue life
assessment methodologies that are based on empirical data can be separated into two categories:


Reliability or scatter analysis
Curve-fit based on flaw growth
Both approaches require a considerable amount of empirical data. However, the first approach
was extended to several programs through the concept of shared databases and in terms of
general scatter of composite data in contrast to metal data. The cumulative effects of data scatter
in different design details of a particular structure are analyzed in the lower levels of the building
block in terms of reliability to determine a safe design service goal or life representative of the
weakest member of the population after a specified life in service. The major limitation in the
second approach is that it is often specific to a certain material system, a loading configuration,
and failure mechanism. As part of the U.S. Navy F/A-18 certification, a probabilistic
methodology was developed to certify composite structures with the same level of confidence as
metallic structures [2]. This methodology was formulated to account for the uncertainties of
applied loads as well as the scatter in static strength and fatigue life related to composite
structures. Over the years, several composite structural certification programs employed this
certification methodology, which was developed for materials and test methods that were
considered current at the time. Since then, materials and process techniques as well as test
methods for evaluating composites have evolved. Consequently, test data often display
significantly less scatter and therefore higher reliability. Thus, the probabilistic approach
employed by Whitehead, et al. [2], can be re-evaluated for newer material forms and to represent
structural details of current aircraft structures to obtain improved life and load-enhancement
factors [3 and 4].
2
Current regulations require airframes to demonstrate adequate static strength, fatigue life, and
damage tolerance capability by testing and/or analysis with a high degree of confidence. These
requirements are intended to account for uncertainties in usage and scatter exhibited by
materials. Analysis is the primary means of structural substantiation for most aircraft
certification programs. It is expected that the analysis is supported by appropriate test evidence.
To develop a certification methodology for composite structures that has the same level of
reliability as observed in metal certification approaches, accounting for the inherent difference
between metal and composites, the FAA and U.S. Navy developed a certification approach for
bolted composite structures [2 and 5] as part of the U.S. Navy F/A-18 certification (figure 2).
This methodology is referred to as NAVY, or the load-life combined approach, throughout this
report.
Figure 2. Material Distribution for U.S. Navy F-18 Aircraft [6]
This approach adopted two key requirements in metallic aircraft certification: (1) the full-scale
static test article must demonstrate a strength that is equal to or exceeds 150 percent of the design
limit load (DLL), and (2) the full-scale fatigue test article must demonstrate a life that is equal to
or exceeds twice the design service life. This approach analyzes the data scatter in the static
strength and fatigue life of composites to establish a certification methodology that has the same
level of reliability as for metal structures. Furthermore, this approach attempts to address the
issues related to hybrid (composite and metallic) structures through a combined approach
referred to as the load-life approach, which will be further discussed in this report. The load-life
approach was developed for what, at that time, was current composite usage and did not
explicitly account for the damage in composite structures or adhesively bonded structural details.
Kan and Whitehead [7] proposed a damage tolerance certification methodology to determine the
reliability of impact damage on a composite structure and to calculate the allowable impact threat
at a given applied load and specified reliability. Subsequent application of this methodology on
3
a U.S. Navy F/A-18 inner-wing structure demonstrated successful damage tolerance capabilities
during certification.
The NAVY load-life methodology was adopted by Shah, et al. [8], for certification of a stiffener
run-out detail. They found that the static-strength and life shape parameters are similar to that
developed for the NAVY approach. This research successfully demonstrated the combined loadlife approach for large-component tests. The applicability of the U.S. Navy damage tolerance
approach of Kan and Whitehead [7] to certification of general and commercial aircraft was
investigated by Kan and Dyer [9]. The Kan and Dyer study showed that the U.S. Navy damagetolerance approach based on military requirements is too severe for the all-composite LearFan
2100 structure.
Early developments of the B-737 graphite/epoxy horizontal stabilizer [10] and the Airbus A310
[11] and A320 [12] all-composite vertical tail used the combined load-life approach for full-scale
demonstrations. The no-growth, damage-tolerant design concept was also adopted whereby a
composite structure must demonstrate the ability to contain intrinsic manufacturing defects and
the maximum allowable service damage(s) in adverse operational conditions and throughout the
design life of the structure. Following the early approach developed for the NASA/B-737
horizontal stabilizer, B-777 empennage certification was primarily based on analysis supported
by coupon and component test evidence [13]. The certification process included general
requirements for environmental effects in design allowables, static strength, and fatigue and
damage tolerance with a no-growth approach. By making predictions prior to testing, such
demonstrations contribute to a solid basis for acceptance of “certification by analysis” by the
FAA and the aviation industry. This is consistent with current certification practices that allow
the use of analysis for certification when supported by tests.
Several all-composite business aircraft, including the Beechcraft 2000 Starship, evolved in the
early 1980s and completed FAA damage tolerance certification requirements [14]. The allcomposite Beechcraft Starship was certified in 1989 using the damage tolerance approach,
identifying environmental effects and concerns related to bonded joints. To meet FAA damage
tolerance requirements, major structural modifications had to be made to the wing. For full-scale
durability and damage tolerance tests, a combined load-life approach based on flaw-growth
threshold stress was employed [15]. The environmental effects were addressed by an analytical
approach validated by testing.
Under the Composite Affordability Initiative, Kan and Kane [16] explored the feasibility of
extending probabilistic methodology for adhesive-bonded composite structures. Three areas
were thoroughly reviewed to determine maturity level: (1) probability theories and probabilistic
methods, (2) probabilistic structural analysis tools, and (3) probabilistic structural criteria and
requirements. The Composite Affordability Initiative identified that the same level of structural
reliability with equivalent level of confidence can be achieved by the probabilistic method
compared to the deterministic method.
Sumich and Kedward [17] investigated the use of the wearout model, on the basis of its
applicability to matrix-dominant failure modes to examine the fatigue performance of the Rotor
Systems Research Aircraft X-wing vehicle. Wearout models assume that structural degradation
4
occurs with use and can be monitored by measuring parameters such as residual strength and
stiffness. Halpin, et al. [18], discussed this methodology in the early 1970s, and several
certification programs, such as the A-7 outer wing and F-16 empennage, have adopted this
methodology for composite structures. This method, which was applied to metal crack growth,
determines fatigue failure when pre-existing damage grows until the specimen can no longer
support the applied cyclic load. In addition, the residual strength of runout is related to crack
length through fracture mechanics. This approach was improved by Sendeckyj [19] using a
deterministic equation that converts static, fatigue, and residual strength data into a pool of
equivalent static-strength data. Sendeckyj’s basic model assumes that the failure in a constantamplitude fatigue test occurs when the residual strength is equal to the maximum cyclic-fatigue
load. This pooling technique for fatigue data is useful for cases where there are not enough
fatigue data in individual stress levels for Weibull analysis, which requires a minimum of six
specimens in each stress level. This model was further improved for pooling fatigue tests with
multiple stress ratios [20], but was not validated since it requires a significant amount of test
data. Stress ratio, or R ratio, is the ratio of minimum-to-maximum cyclic stress in a fatigue test.
O’Brien and Reifsnider [21] studied fatigue life analytically using the fatigue modulus concept.
This approach assumed that fatigue failure occurs when the fatigue secant modulus (residual
stiffness) degrades to the secant modulus at the moment of failure in a static test. In this study,
stiffness reductions resulting from fatigue damage were measured for unnotched [±45]s, [0/90]s,
and [0/90/±45]s boron/epoxy laminates. Degradation in the various in-plane stiffnesses (axial,
shear, and bending) was measured using a combination of uniaxial tension, rail shear, and
flexure tests. Damage growth and stiffness loss were identified to be load-history dependent.
Hence, the secant modulus criterion was not a valid criterion for general applications. A similar
study was conducted on the fatigue behavior of [0/±45/90]s glass/epoxy laminate by Hahn and
Kim [22] in which the secant modulus was used as a measure of damage extent.
Following an extensive review of different damage models, Hwang and Han [23] identified
various cumulative damage models using several physical variables, such as fatigue modulus and
resultant strain. They introduced a new concept called “fatigue modulus,” which is defined as
the slope of applied stress and resultant strain at a specific cycle [24]. Fatigue modulus
degradation assumes that the fatigue modulus degradation rate follows a power function of the
fatigue cycle. The theoretical equation for predicting fatigue life is formulated using the fatigue
modulus and its degradation rate. This relation is simplified by the strain failure criterion for
practical applications. Mahfuz, et al. [25], analytically studied the fatigue life of an
S2-glass/vinyl-ester composite using the fatigue modulus concept. This study revealed that the
fatigue modulus is not only a function of loading cycle but also a function of applied stress level
and thickness of the test specimen. This life-prediction methodology requires two parameters
that are obtained empirically either at two different stress levels or two different fatigue
lifetimes.
Halpin, et al. [26], suggested that the fatigue behavior of composites should be based empirically
under particular design spectra. The main disadvantage of such an approach is that test results
are specific to a loading spectrum. Also, a large number of test data is required for a complete
analysis, like the extensive fatigue sensitivity study conducted by Jeans, et al. [27], on bolted and
bonded composite joints under various loading spectra. For metals, Miner’s rule is often used to
5
study the cumulative damage under a loading spectrum. However, Rosenfeld and Huang [28]
conducted a fatigue study with different stress ratios to determine the failure mechanisms under
compression of graphite/epoxy laminates and showed that Miner’s rule fails to predict composite
fatigue under spectrum loading. This is confirmed by several authors in the composite
community. A study conducted by Agarwal and James [29] on the effects of stress levels on
fatigue of composites confirmed that the stress ratio had a strong influence on the fatigue life of
composites. Further, they showed that microscopic matrix cracks are observed prior to gross
failure of composites under both static and cyclic loading.
For practical consideration, Yang and Du [30] investigated the possibility of statistically
predicting the fatigue behavior of composites under service-loading spectra, based on some
baseline constant-amplitude fatigue data. Although such a phenomenological statistical model
does not account for the intrinsic failure mechanisms that are quite complex in composite
materials, it can be very simple for practical applications and requires significantly less empirical
effort.
Kassapoglou [31] presented a probabilistic approach for determining fatigue life for composite
structures under constant-amplitude loading. This approach assumes that the probability of
failure during any cycle is constant and equal to the probability of failure obtained from static
test results and associated statistically quantified scatter. This methodology does not require any
fatigue data for calibration or for the expression of the cycles to failure as a function of stress
ratio. Comparison of fatigue life predictions for several stress ratios with a number of
experimental data shows good correlation. However, the assumptions used in this model neglect
the complex progressive damage mechanism that takes place during repeated loading.
1.1 BACKGROUND FOR CURRENT APPROACH.
The current practice is to test composite structures with loads that are enhanced to reduce long
test duration requirements. This accounts for the data scatter observed in composites relative to
metals. The load-life approach proposed for U.S. Navy F/A-18 certification [2] is used such that
the same level of reliability as for metal structures can be achieved. Compared to the metal static
and fatigue data, composite materials exhibit high data scatter due to their anisotropic
heterogeneous characteristics, such as lay-up, manufacturing defects and imperfections, test
complications, and environment. To interpret the information in a meaningful manner and to
incorporate any of their effects into the certification of composite structures, the life-factor
approach and the load-enhancement factor (LEF) approach are commonly used and require
composite scatter analysis, which is described in this report. The life-factor approach, which has
been successfully used for metallic structures to assure structural durability, accounts for the
scatter in life (S/N) data in terms of the shape parameter of the population. The life shape
parameter (often referred to as the modal life shape parameter) is obtained by analyzing the
distribution of the shape parameters corresponding to S/N curves representing different design
details of the structure. The life factor corresponds to the relation between the central tendency
(mean) of the population and the extreme statistics (allowable). The underlying objective of the
life-factor approach is to ensure that the design service goal or life is representative of the
weakest member of the population after a specified life in service. Thus, a successful repeated
load test to mean fatigue life would demonstrate the B-basis reliability on the design lifetime.
The NAVY approach showed that the life shape parameters of metal and composite are 4.00 and
6
1.25, respectively, and they correspond to life factors of 2.093 and 13.558, respectively, for
B-basis reliability [2]. Therefore, due to the large scatter in composite test data, a composite
structure is required to test additional fatigue life to achieve the desired level of reliability, i.e., a
test duration of more than 13 design lifetimes is required for composite in contrast to two design
lifetimes for metal to achieve B-basis reliability.
An alternative approach to the life factor, which requires an excessive test duration, is to increase
the applied loads in the fatigue spectrum so that the same level of reliability can be achieved with
a shorter test duration [2]. This approach is referred to as the LEF approach and was derived
from combining the life factor and the static factor (ratio of mean-to-allowable fatigue strength)
at one lifetime to form a relationship between the LEF and the test duration. The static factor is
defined in terms of a static-strength shape parameter that is obtained by analyzing the
distribution of the shape parameters corresponding to static-strength data sets representing
different design details of the structure as described in section 4. The formal relationship
between the LEF and the test duration provides the flexibility of conducting the durability test of
a composite structure with different LEFs and corresponding test durations to achieve the desired
reliability. Although the materials, processes, lay-up, loading modes, failure modes, etc., are
significantly different, most current certification programs use the load-life factors generated for
the U.S. Navy F/A-18 certification program. Lameris [3] showed that both LEFs and life factors
can be significantly reduced by using strength and life shape parameters generated for materials,
processes, loading modes, failure modes, etc., applicable to a specific structure. However,
guidance for developing these shape parameters is greatly needed.
Although fatigue life is adversely affected by damage (notch), the scatter in damaged
composites, both in static and fatigue, tends to decrease due to localized stress concentration.
This will result in lower life and load-enhancement factors. Therefore, scatter analysis of
coupons/elements in lower levels of building blocks can be used to develop a synergy among the
life factor, LEF, and damage in composites. This approach is beneficial for the damage tolerance
phase of full-scale substantiation and minimizes the risks associated with the introduction of
large damage to durability test articles.
Development of scatter analysis applicable to current composite materials and processes using
improved test methodologies demonstrate lower requirements for the life factor and LEFs.
Introduction of damage philosophy into the scatter analysis further reduces these factors. The
probabilistic approach employed in the NAVY load-life combined approach shows the potential
use of improved shape parameters for estimating the effects of design changes, i.e., gross weight
changes, on design life. This requires a probabilistic approach to redefine basis (A- or B-basis)
fatigue life requirements set forth in the load-life combined approach to any deviation from the
life (i.e., reduction in life factor due to damage introduction) or load factor (i.e., high spectrum
fatigue loads due to gross weight change). For a full-scale test that is conducted using a higher
LEF or is completed more than the required test duration, this technique can be used to redefine
original design service goals (number of hours equivalent to one life) associated with the fatigue
spectrum.
7
1.2 RESEARCH OBJECTIVES AND OVERVIEW.
The research methodology discussed here consists of combining existing certification approaches
utilized by various aircraft manufacturers with protocols for applying these methodologies. This
will extend the methodologies to new material systems and construction techniques. This report
includes data for materials commonly used in aircraft applications, including adhesives and
sandwich construction. The testing consisted of various element-type tests and concentrated on
tests that were generic in nature and representative of various loading modes and construction
techniques. Three different techniques are discussed for scatter analysis of fatigue data:
individual Weibull, joint Weibull, and the Sendeckyj wearout model. Procedures to generate
reliable and economical scatter and LEFs necessary for a particular structural test by selecting
the design details representing the critical areas of the structure are outlined with several
examples and case studies. The effects of laminate stacking sequence, test environment, stress
ratios, and several design features, such as sandwich and bonded joints, on the static-strength and
fatigue shape parameters are discussed with detailed examples. Furthermore, several analytical
techniques for obtaining these shape parameters are discussed with examples. Finally, the
application of LEFs and life factors on a full-scale test spectrum without adversely affecting the
fatigue life and the damage mechanism of the composite structure is discussed.
The data from this report describes the first phase of a research program outlined in figure 3.
The key objective of this research was to develop a probabilistic approach to synthesizing the life
factor, LEF, and damage in composite structures to determine the fatigue life of a damagetolerant aircraft. This methodology was extended to the current certification approach to explore
extremely improbable high-energy impact threats, i.e., damages that reduce the residual strength
of aircraft to limit-load capability and allow incorporating certain design changes into full-scale
substantiation without the burden of additional time-consuming and costly tests. Research was
conducted in three phases (figure 3):



Load-life combined approach
Damage tolerance and flaw-growth tests
Load-life damage (LLD) hybrid approach
The first subtask of this research phase was intended to generate a database of fatigue life data
for several composite material systems that are commonly used in general aviation. The second
subtask in this phase was to add static-strength shape parameters to the database and generate
improved LEFs for several example materials. These data were then used to generate necessary
load-life combined data, for example, full-scale demonstrations included in the final stage of the
research. The improvements in materials and processes and test methods produced life factors
and LEFs lower than the values commonly used in most certification programs based on the
NAVY approach. Data gathered in this phase were used to provide guidance for generating safe
and reliable LEFs and life factors pertaining to a specific structure. In addition, a user-friendly
computer code that can be used for scatter analysis of composites was developed. This code
alleviates misinterpretation of any statistical or mathematical processes during the analysis and
provides guidance for selecting different techniques appropriate for a particular application. This
report only contains the data generated during the first phase of the research.
8
FEA = Finite element analysis
ADL = Allowable damage limit
CDT = Critical damage threshold
RDL = Repairable damage limit
Figure 3. Overview of Research
2. EXPERIMENTAL PROCEDURE.
The primary goal in the first phase of the research was to interrogate the methodology for the
development of Weibull parameters to be used in the load-life combined approach. Two key
parameters were needed: static-strength and the fatigue-life shape parameters. The first subtask
in this phase was to investigate the life factor for several composite material systems using the
fatigue shape parameter. Then, using fatigue-life and static-strength shape parameters, the LEF
for several material systems was calculated. Finally, a comparison of the load-life approach for
several material systems and design scenarios was shown with two benchmark case studies: the
Beechcraft Starship forward wing and Liberty XL2 fuselage. The second phase incorporated
different damage categories into a full-scale test article and investigated the effects of damages
on life factors and LEFs. The final phase was intended to develop a hybrid approach using the
life factor, LEF, and damage in the composite. Once the load-life factors were generated for the
Beechcraft Starship material, full-scale fatigue tests of the last phase were performed to verify
the LLD approach.
9
2.1 MATERIAL SYSTEMS.
The three main material systems studied for the purpose of generating static and fatigue shape
parameters were Cytec AS4/E7K8 plain-weave fabric (AS4-PW), Toray T700SC-12K50C/#2510 plain-weave fabric (T700-PW), and 7781/#2510 8-Harness glass-fiber fabric (77818HS) [32]. The test data for these three materials are referred to as FAA-LEF data throughout
this report. In addition to the FAA-LEF data, a detailed static scatter analysis was conducted on
the following materials available through an extensive laminate database [33]: Toray T700G12K-31E/#2510 unidirectional tape and T700SC-12K-50C/#2510 plain-weave fabric, Advanced
Composites Group (ACG) MTM45/AS4C 12K unidirectional tape and MTM45/AS4C 6K
5-harness graphite fabric, and Nelcote® (formally FiberCote) T700-24K/E765 unidirectional tape
and T300-3K/E765 plain-weave fabric. This data set is referred to as the FAA lamina variability
method (FAA-LVM) data throughout this report and was generated to analyze the LVM [34] on
generating laminate allowables.
In addition to the above two data sets, fatigue scatter analysis for Loctite, Hysol EA9696, and
PTM&W ES6292 adhesive systems were included from the data obtained from FAA research to
investigate the durability of adhesive joints [35]. These test specimens were fabricated according
to an ASTM standard test method for thick-adhered metal lap-shear joints to determine the shear
stress-strain behavior of adhesives in shear by tension (ASTM D 5656). This data set is referred
to as FAA-D5656 data throughout this report. Finally, element test data of adhesively bonded
composite joints that were loaded in picture-frame shear (PFS) and single-lap shear (SLS) [36]
test configurations were included in the analysis. Data from this database are referred to as the
FAA effects of defects (FAA-EOD) data throughout this report. Once the scatter analysis was
completed, LEFs were generated, combining scatter analysis of the above data sets for available
fatigue cases.
The Beechcraft Starship was primarily fabricated using an AS4/E7K8 epoxy material system
(original manufacturer: U.S. Polymeric). Hercules AS4 fibers are continuous carbon filaments
made from a PAN precursor, and their surface is treated to improve handling characteristics and
structural properties. Typical fiber tensile modulus and strength are 34 Msi and 550 ksi [37].
The E7K8 medium-flow epoxy resin system has good tack characteristics for handling and a
20-day out-time at ambient temperature. The AS4/E7K8 3K plain weave material system
(AS4-PW) has an aerial weight of 195 g/m2, a typical cured-ply thickness of 0.0087 inch, and a
low-exotherm profile for processing thick parts. This material is currently being used by Hawker
Beechcraft and Cessna Aircraft in Wichita, Kansas, for several aircraft applications.
2.2 TEST MATRICES.
Testing included various element-type tests and concentrated on generic tests that would be
representative of various loading modes and construction techniques. In general, this program
primarily focused on stress ratios within the wing and fuselage envelopes for the development of
the Weibull fatigue shape parameter. Using the data gathered in the lamina, laminate, and
element tests, the methodology used to develop the Weibull static-strength parameters was
compared for various scenarios.
10
Commonly used laminate stacking sequences—hard (50/40/10 for unidirectional tape and
40/20/40 for fabric), quasi-isotropic (25/50/25), and soft (10/80/10) laminate constructions—
were used for the FAA-LEF database (table 1).
Table 1. Laminate Configurations for FAA-LEF Database
Laminate
Hard
Quasi-isotropic
Lay-Up %
0°/45°/90°
40/20/40 (weave)
25/50/25
Soft
10/80/10
All ±45
0/100/0
Ply Stacking Sequence
[0/90/0/90/45/-45/90/0/90/0]S
[(45/0/-45/90)2]S
[(45/0/-45/90)4]S
[45/-45/90/45/-45/45/-45/0/45/-45]S
[45/-45/90/45/-45/45/-45/0/45/-45]2S
[(45/-45)5]S
Total
Plies
20
16
32
20
40
20
In addition, sandwich specimens were fabricated with three-ply facesheets (plies in the 0°
direction) with a 0.25-inch-thick HRH-10 Nomex core. Test methods and fixture requirements
for FAA-LEF tests are shown in table 2. Although these test methods are recommended for
static testing, a similar test setup was used for fatigue tests. Double-notched compression (DNC)
specimens were modified to have similar geometry to open-hole compression (OHC) (12 x 1.5
inches), and the OHC fixture was used for both static and fatigue testing. Similarly,
ASTM D 3165 specimens were modified to have the same overall specimen dimensions with a
1.5-inch overlap so the OHC fixture could be used for compression loading.
Table 2. The FAA-LEF Test Methods and Fixture Requirements
Test Description
Tension, open hole
Compression, open hole
Double-notched compression
Single-lap shear, tension
Single-lap shear, compression
Sandwich four-point bend
Compression after impact
Tension after impact
Abbreviation
OHT
OHC
DNC
SLS-T
SLS-C
4PB
CAI
TAI
11
Test Method
ASTM D 5766
ASTM D 6484
Modified ASTM D 3846
Modified ASTM D 3165
Modified ASTM D 3165
ASTM C 393
ASTM D 7137
Modified ASTM D 3518
Test
Fixture
No
Yes
Yes
No
Yes
Yes
Yes
No
The basic FAA-LEF test matrix is shown in table 3. All 10/80/10 laminates in this table were
fabricated with a 20-ply stacking sequence, as shown in table 1. To support the full-scale
demonstration and the damage tolerance efforts shown in figure 3, a supplemental test matrix
was added for AS4-PW (table 4). These test matrices represent different lay-ups, test
environments, loading modes, bonded joints, and sandwich structures. Sandwich specimens
were fabricated with HexWeb® HRH-10 manufactured from Nomex® aramid fiber sheets. This
core was selected because of its application in the Beechcraft Starship.
Table 3. The Basic FAA-LEF Test Matrix
Laminate
10/80/10
Laminate
Test
Method
OH
SLS
(t = 0.01″)
SLS
(t = 0.06″)
DNC
Sandwich
4PB
Loading
Condition
Standard
Static Test
Environment
RTA—Cyclic Test R-Ratio
(Three Stress Levels)
ETW
6
-0.2
18
Tension
ASTM D 5766
RTA
6
Compression
ASTM D 6484
6
6
Adhesive
In-Plane
Shear
Modified
ASTM D 3165
6
6
6
6
Interlaminar
Shear
Modified
ASTM D 3846
6
6
Flexure
Modified
ASTM C 393
6
6
0
18
-1
5
18
18
18
18
18
RTA = Room temperature ambient
ETW = Elevated temperature 180°F, wet
OH = Open hole
The 10/80/10 CAI specimens in table 4 were fabricated with the 40-ply stacking sequence, while
the 25/50/25 CAI specimens were fabricated using the 32-ply stacking sequence, as shown in
table 1. The 25/50/25 CAI specimens were machined to 6 inches by 9 inches to minimize the
edge effects for larger damages and to leave room for damage propagation during cyclic loading.
Extensive testing of coupons [38] and components [39] of adhesive joints has shown a
significant decrease in static strength for thick bondlines. Thus, the adhesive joints with different
bondline thicknesses were included in the test matrix. ASTM D 3518 was modified to have a
4-inch width for the TAI specimen, with impact at the center.
12
Table 4. Supplemental FAA-LEF Test Matrix
Laminate
10/80/10
Laminate
25/50/25
Laminate
Test
Method
CAI
OH
CAI
40/20/40
Laminate
0/100/0
Laminate
CAI
OH
TAI
Loading
Condition
Compression
BVID
Compression
VID
Compression
RTA
Compression
ETW
Unimpacted
RTA
Compression
BVID/RTA
Compression
VID/RTA
Compression
LID/RTA
Compression
VID
Compression
RTA
Shear
BVID/RTA
Shear
VID/RTA
Standard
ASTM D 7137
Static Test
Environment
Cyclic Test R-Ratio
(Three Stress Levels)
RTA
6
-0.2
ETW
0
-1
6
ASTM D 6484
5
18
18
6
18
6
ASTM D 7137
6
6
18
6
18
6
18
ASTM D 7137
6
18
ASTM D 6484
6
Modified
ASTM D 3518
6
18
6
18
18
RTA = Room temperature ambient
ETW = Elevated temperature 180°F, wet
LID = Large impact damage
2.3 EXPERIMENTAL SETUP.
This section contains information regarding the experimental setup and equipment used for
impact, nondestructive inspection (NDI), and residual strength tests of the FAA-LEF test
specimens.
2.3.1 Impact Tests.
CAI and TAI test specimens were impacted using an Instron Dynatup 8250 drop-weight impact
tester (figure 4). The impact force was measured using a piezoelectric load cell attached to the
impact mass assembly. This impact tester was equipped with a pneumatic rebound catch
13
mechanism, which prevents secondary impacts on the test specimens, and a photo-detector/flag
system, which provides impact velocity information. Data acquisition software, which runs on a
computer connected to the drop-weight impact tester, collected and reduced the impact test data.
A sensor (flag), which was placed close to the impact location, triggered the data acquisition
system a few milliseconds prior to the impact event. This sensor and another flag placed a
known distance away were used for calculating impact velocity (velocity = distance/time).
Prior to impacting, specimens were placed in the support fixtures, as shown in figure 5, and held
rigidly. These fixtures use dowel pins for aligning the specimens. The total impact event
duration was at most 10 milliseconds. Therefore, a very high-frequency triggering mechanism,
an Instron Dynatup impulse data acquisition system, was used to collect data during the impact
event.
Figure 4. Instron Dynatup Drop-Weight Tester
Figure 5. Support Fixture for CAI and TAI Impact Specimens
14
2.3.2 Nondestructive Inspections.
Impacted specimens were subjected to through-transmission ultrasonic (TTU) NDI that
generated C-scans to quantify the planar-damaged areas using image analysis software (figure 6).
Additional inspections techniques, e.g., microscopy and thermal imaging, were also used for the
damage tolerance investigation. For those cases involving glass fiber composite, damage can be
seen clearly with the naked eye due to the translucent nature of these fibers.
Figure 6. The TTU Scanning of PFS Test Specimen
In addition to TTU C-scans, test specimens were inspected with the Sonic 1200 ultrasonic flaw
detector and BondMasterTM 1000 hand-held NDI inspection units while the specimens were in
the test setup. The BondMasterTM 1000 is capable of resonance, mechanical impedance analysis
(MIA), and pitch/catch mode, and the user has the ability to select the method best suited for
inspecting a particular composite structure. The MIA technique, which was used for inspecting
test specimens in this program, measures the stiffness and mass of the material under test and
requires no coupling agents. The output was measured in both amplitude and phase. Both of
these hand-held units are equipped with color displays and provide real-time data.
2.3.3 Full-Field Strain Evolution.
The ARAMIS photogrammetry full-field strain measurement system (figure 7) was used to
measure localized buckling in the region of disbonds/defects. ARAMIS [40] is a noncontact,
optical, three-dimensional deformation measuring system. It uses two high-definition cameras to
track translation and rotation of the surface details (object characteristics) with subpixel
accuracy. Surface details are obtained by applying a stochastic color pattern that follows surface
displacement during loading. ARAMIS uses this pattern to recognize the surface structure and
then uses digitized images from both cameras for triangulation of surface details (micro-pattern)
to determine the precise location of each point. Therefore, this system has the capability of
digitizing the precise shape (surface) of the structure during loading. The first set of coordinates
for object characteristics is obtained in the undeformed stage. After load application, a new set
of coordinates (digital images) is recorded. Then, ARAMIS compares the digital images and
calculates the displacement and deformation of the object characteristics.
15
Figure 7. Portable Version of ARAMIS Photogrammetry System [40]
ARAMIS is capable of three-dimensional deformation measurements under static and dynamic
load conditions to analyze deformations and the strain of real components. In addition, this
system is able to eliminate the rigid-body motion component from the displacement results.
Therefore, it can be used for specimens that exhibit large displacements. Strain sensitivity of the
system is approximately 100-200 microstrains, and the scan area can be as large as 47 inches by
47 inches. Full-field displacement/strain data are then used to examine any propagation of the
defects according to the procedures outlined by Tomblin, et al. [41], which assess the localized
skin buckling (out-of-plane displacement) around the disbonded or delaminated region. The
full-field strain evaluation of CAI specimens during static (figure 8) and fatigue loading was
measured using the ARAMIS photogrammetry system.
40000
Failure
35000
Compressive Load (lbf)
30000
25000
20000
15000
Strain 1
Strain 2
10000
Strain 3
Strain 4
5000
0
0
1000
2000
3000
4000
5000
6000
7000
8000
Far-Field Strain (microstrain)
Figure 8. Damage Evolution of a CAI Specimen Under Static Loading
16
2.3.4 Static and Residual Strength Tests.
All static and residual strength tests were conducted using Material Test Systems (MTS)
servohydraulic test frames. Test specimens that did not require fixtures were mounted to the test
frame using a hydraulic grip assembly, as shown in figure 9. While gripping the specimens, the
actuator was programmed in load-control mode to prevent unnecessary preloading due to grip
pressure. Static tests were conducted in displacement-control mode at a rate of 0.05 in/min while
acquiring data at a rate of 10 Hz.
Figure 9. The MTS Servohydraulic Test Frame
2.3.5 Fatigue Life Evaluation.
Fatigue tests were conducted in load-control mode at a frequency of 5 Hz. Fatigue specimens
included several R-ratios that represented loading levels in different parts of the aircraft
(section 2.2). Fatigue tests were conducted at three different stress levels with a minimum of six
specimens per stress level to support the minimum requirements of individual Weibull analysis.
Specimen compliance degradation was monitored throughout fatigue duration to examine
damage evolution.
To investigate the damage progression, full-field strain data were interpreted according to the
NDI method outlined by Tomblin, et al. [41]. Both ARAMIS and C-scan data were used to
establish guidelines for determining the fatigue failure of specimens that was not obvious, e.g., a
four-point bend sandwich specimen did not indicate any sign of complete delamination across
the width and continued to hold applied cyclic loading (figure 10).
17
Figure 10. Compliance Change and Damage Area During Fatigue Tests of
4PB Sandwich Specimen
Antibuckling fixtures were used for compression fatigue specimens, such as open hole (OH),
DNC, and CAI, to prevent premature failure. The OHC fixture, which is designed for static
tests, indicated wear as the specimen compliance changed during the fatigue tests and required
modifications to prevent further damage to the fixture and load misalignment during the fatigue
tests. The change in temperature was monitored for several specimens with defects and was
found to be insignificant, i.e., less than 10°F.
Several trial specimens were used at the beginning of each loading mode and lay-up to determine
appropriate stress levels so that at least two provided fatigue failures. Fatigue loads for each
specimen were calculated with respect to static strength using the actual specimen dimensions.
3. ANALYSIS OF COMPOSITE TEST DATA.
Compared to metal static and fatigue data, composite materials exhibited higher data scatter due
to their anisotropic heterogeneous characteristics, such as lay-up, manufacturing defects and
imperfections, test complications, and environment (figure 11). To interpret this information in a
meaningful manner and to incorporate any effects of this into the certification of composite
18
structures, several approaches were used. Life factor, LEF, and the load-life combined approach
are three of the commonly used approaches that require composite scatter analysis.
Figure 11. Life Scatter in Composites and Metal [2]
3.1 SCATTER ANALYSIS.
The primary goal in scatter analysis of composites is to interpret the variability in data in lower
levels of the building blocks of testing and translate the statistical significance of such
phenomenon into full-scale test substantiation. To determine the shape parameters for static
strength and fatigue life for the purpose of full-scale test substantiation, test matrices must be
designed so that at least the design details and loading modes of critical locations of the structure
are represented by coupon and/or element tests. The influence of material, lay-up sequence,
loading mode, sandwich construction, joints, environmental effects, etc., is typically considered
during static-strength scatter analysis. Fatigue analysis includes the influence of the stress ratio
in addition to the above-mentioned design details.
Scatter in composites can be analyzed as Weibull distribution or normal distribution. Since the
shape parameter provides information with respect to the data scatter, the two-parameter Weibull
distribution is commonly used in composite static and fatigue scatter analyses. Fatigue scatter in
composite test data can be analyzed using several different techniques. These techniques are
mainly subdivided into two categories: individual analysis and pooling. Joint Weibull and
Sendeckyj are the two pooling techniques discussed in this report.
First, the shape parameters corresponding to the different data sets representing different design
detail (denoted by αˆ ) are obtained using Weibull analysis; then, the shape and scale parameters,
and, corresponding to the Weibull distribution of α̂ ’s are used to calculate the modal or
mean α̂ , which is referred to as the static-strength or fatigue-life shape parameter. To be
conservative, the modal value of the distribution of shape parameters is selected as the strength
or life shape parameter, rather than the mean value (figure 12), and this is referred to as the
modal static-strength shape parameter (MSSP) or modal fatigue-life shape parameter (MLSP),
respectively.
19
Determination of fatigue life scatter requires a large number of test replications at different stress
levels. To reduce cost and test duration while maintaining the reliability of data analysis, it is
recommended that the fatigue scatter analysis be conducted using pooling methods such as the
joint Weibull or Sendeckyj wearout analysis.
Frequency
of
Occurrence
Modal α̂
Mean αˆ
α and β - Shape and scale
parameters of the Weibull
distribution of αˆ
Figure 12. Scatter Analysis Using Weibull Distribution of Shape Parameters
3.1.1 Individual Weibull Method.
Weibull distribution is used in statistical analysis of composites, especially for small samples,
due to its simple functionality and ease of interpretation. The commonly used two-parameter
Weibull distribution expressed by the cumulative survival probability function is shown as
ˆ
ˆ
P ( X  x  e)  ( x /β)
(1)
where, x is the random variable, α̂ is the shape parameter, and β̂ is the scale parameter.
The population mean, , and standard deviation, , are calculated with the Gamma ()
distribution function in equations 2 and 3, respectively.
 α̂  1 
μ  β

 α̂ 
(2)
 αˆ  2 
 αˆ  1 
 2 
σ  βˆ  


 αˆ 
 αˆ 
(3)
The shape and scale parameters are estimated in an iterative process using either the maximum
likelihood estimation (MLE) or rank regression [42]. Rank regression in X (RRX) tends to
produce reliable results for small samples, while MLE works well for samples containing more
than 20 or 30 data points.
During the individual Weibull analysis of fatigue data, each stress level is analyzed, and then the
shape parameters are arithmetically averaged to define life scatter. Since Weibull analysis
20
considers only the data at a certain stress level at a time, five or more data points must be
included in each stress level. For S/N data that has less than five data points per stress level,
either joint Weibull analysis or Sendeckyj analysis must be used.
3.1.2 Joint Weibull Method.
In the joint Weibull analysis, M groups of data having a common shape parameter, but different
scale parameters, are pooled [43]. The common shape parameter is obtained using the joint
maximum likelihood estimate method, as shown in equation 4a. For the special case, where
there is equal number of failures across all stress levels (i.e., nf1, nf2,.., nfi = nf), equation 4a reduces
to the form shown in equation 4b.
n fi

 M α̂

x
1
n
x
1
n
(
x
)









ij
ij
ij
M
1 j 1


j 1


0
 
n fi  

ni
 
α̂
n fi
I 1 
αˆ
xij




1

j



 n fi

 ni α̂


x
1
n
(
x
)
1
n
(
x
)




ij
ij
ij


M
M M  j 1
j 1
0


 

ni




α̂
n
i 1
i 1
fi
xijαˆ





j 1




(4a)
4(b)
where ni = number of data points in the ith group of data (i = 1,2,…., M)
nfi = number of failures in the ith group of data (i = 1,2,…., M)
The common scale parameter for the S/N data is obtained using equation 5.
1
 1 ni

βi     xijα̂ 
 n j 1 
 fi

α̂
(5)
3.1.3 Sendeckyj Equivalent Static-Strength Model.
The Sendeckyj equivalent static-strength (wearout) model [19] uniquely relates the static
strength and residual strength to fatigue life. Thus, the analysis pools static strength, fatigue life,
and residual static-strength data and converts it into equivalent static-strength data. To
characterize the S/N behavior of composite materials, the Sendeckyj model assumes:

The S/N behavior can be determined by a deterministic equation. The equation can be
based on theoretical considerations or experimental observations of the fatigue damage
accumulation process.
21

The static strengths are uniquely related to the fatigue lives and residual strengths at
runouts. This specific relationship assumes that the strongest specimen has the longest
fatigue life or the highest residual strength at runout; fatigue failure is assumed when the
residual strength is equal to the maximum amplitude cyclic stress. This assumption may
not hold for cases where competing failure modes are observed during fatigue testing.

The statistical variability of static-strength data can be described by a two-parameter
Weibull distribution.
The basic Sendeckyj model is presented in the deterministic equation given by
 σ 1
σ e  σ α  r 
 σ α 

S

 (n f  1 C 


S
(6)
where e is the equivalent static strength, a is the maximum applied cyclic stress, r is the
residual strength, nf is the number of fatigue cycles, and S and C are Sendeckyj fitting
parameters. During scatter analysis of fatigue data, these parameters are recalculated for each
fatigue data set (S/N curve) and are used for developing fitting curve for each S/N curve. Setting
the maximum amplitude cyclic stress equal to residual strength for fatigue failures, the power
law is obtained as
σ α  (1  C  C  n f ) S  σ u (7)
where u is the static strength.
Using the Sendeckyj analysis, fatigue life and residual strength data for each S/N curve are
converted into a pool of equivalent static-strength data points. Then, this data set is fitted into a
Weibull distribution to obtain the life shape parameter as described by Sendeckyj [19].
3.2 LIFE-FACTOR APPROACH.
The life-factor approach has been successfully used for metal to assure structural durability. In
this approach, the structure is tested for additional fatigue cycles to achieve the desired level of
reliability. This is graphically illustrated in figure 13 in terms of B-basis statistics, i.e.,
successful repeated load test to mean fatigue life demonstrates B-basis reliability on design
lifetime. Since the true population distribution of life is not readily available, a shape parameter
obtained from fatigue life data used for design from different levels in the building-blocks of
testing is used for representing this probability density function, i.e., fatigue-life shape parameter
in section 5.
The ratio of the mean repeated load life to A- or B-basis repeated life is defined as life factor, NF,
and given by equation 8. The derivation of the general form of this equation is included in
Whitehead, et al. [2].
22
90% of population
Frequency of
Occurrence
True Population
Distribution
Test Duration
Design Life
(B-basis)
Mean Life
Life
Figure 13. Life-Factor Approach for Substantiating B-Basis Design Life
NF 
 α 1 
 L

 αL 




 1n( R) 
 2

   γ (2n)  
2n  



1
(8)
αL
where L is the modal fatigue-life shape parameter (MLSP), n is the number of articles, and R is
the reliability. For γ = 0.95, A- and B-basis reliabilities are 0.99 and 0.90, respectively. (2n)
is the Chi-square distribution with 2n degrees of freedom at -level confidence.
Figure 14 shows the influence of the shape parameter on the life factor, which is the ratio
between mean repeated life to B-basis life [2]. For small , life factor reduces with the
increasing number of test articles. This figure shows that the life factor rapidly increases for
fatigue shape parameters that are less than 2. Due to large scatter in the composite test data, the
life shape parameter of composite was found to be 1.25 for the data analyzed for the NAVY
approach, while it was found to be 4.00 for metal. Therefore, a composite structure is required to
test additional fatigue life to achieve the desired level of reliability, i.e., a test duration of more
than 13 design lifetimes (DLT) is required for composite in contrast to 2 DLT for metal. As
shown in equation 8, life factor is a function of MLSP. Thus, improvements in MLSP for newer
forms of materials that exhibit less scatter can significantly reduce the life factor.
23
20
n=1
n=5
Composites Alpha = 1.25
13.558
9.143
7.625
Metals Alpha = 4.0
2.093
1.851
1.749
n=15
n=5
n=1
Composte [ =1.25]
Metal [ =4.00]
15
Life Factor
n = 15
10
5
0
0.00
1.00
2.00
3.00
4.00
5.00
Wiebull Shape Parameter ( )
Figure 14. Influence of Fatigue Shape Parameter on B-Basis Life Factor
Furthermore, the analysis in reference 4 shows that the data scatter of notched or damaged
composite elements can be significantly less than that of the unnotched composite specimens.
Such improvements in fatigue-life shape parameter can significantly reduce the life factor.
However, the life factor becomes insensitive to small changes in the life shape parameter beyond
a value of 4, which is considered to be the life shape parameter for metal. The composite MLSP
of 1.25, which was used for the NAVY approach, lies within the highly sensitive region of life
factor versus shape parameter curve (figure 14); thus, even a small improvement resulted in a
dramatic reduction of life factor, which reflects the required number of test durations to achieve a
certain level of reliability in the design life. The MLSP is obtained from a distribution of shape
parameters representing numerous S/N curves of different critical structural details. Thus, it is
common to have large scatter in S/N data of design details that have competing failure modes
and less scatter in notched test data due to stress concentration. For example, a V-notched, rail
shear (VNRS) test specimen that has a soft (10/80/10) laminate stacking sequence has a majority
of its fibers aligned with the tensile and compressive load resultant axes during in-plane shear
loading. Although the applied (external) load is an in-plane shear load, the tensile and
compressive (internal) loads along the fiber directions often cause fiber breakage and buckling,
respectively, and significantly contribute to the final failure. In some cases, the competing
(tensile and compressive) loading configurations result in unacceptable failure modes of these inplane shear specimens. Often, the complex state of stress and these competing failure modes,
coupled with other variabilities associated with composites such as batch variability, porosity,
and fiber misalignments, tend to cause large scatter in both static strength and fatigue life. On
the other hand, the stress concentrations in notched composites cause the final failure of the
specimen, negating or minimizing the collective effects of the above-mentioned secondary
variables.
24
3.3 LOAD-FACTOR APPROACH.
LEF is an alternative approach to the life-factor approach, which requires an excessive test
duration and increases the applied loads in the fatigue spectrum so that the same level of
reliability can be achieved with a shorter test duration [2]. A formal relationship between LEF
and the test duration, N, as shown in equation 9, is defined for composite structural certification
[2].
LEF ( N ) 
 α 1
 L 
 αL 
αL

 1n( R)  N α L
 2
  γ (2n) 
2n 
 
αR





1
(9)
αR
Tables 5 and 6 include A- and B-basis LEFs, respectively, using equation 9 for R =17.5 to 32.5
and L=1.25 to 2.50. In these tables, the LEFs are calculated for combinations of test duration,
N, ranging from 1 through 5 design lifetimes, and for 1, 2, and 3 test articles.
LEF for a test duration of 1 DLT, referred to as load factor (LF), is calculated as
LF  λ 
 α 1 
 R

 αR 




 1n( R) 
 2

   γ (2n)  
2n  
 

1
(10)
αR
where, R is the MSSP, and  is a function of both MSSP and MLSP, as defined in equation 11.
Equation 10 has the same form as equation 8, except L is replaced by R, and the new parameter
 is included so that both life factor and LEF have the same level of reliability.
αL
 α  1  αR
 L

αL 

λ
 α 1
 R

 αR 
25
(11)
Table 5. A-Basis LEFs
1
n
N
1
1.5
2
26
3
4
5
L
R
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
3
2
1.00
1.25
1.50
1.75
2.00
2.25
2.50
1.00
1.25
1.50
1.75
2.00
2.25
2.50
1.00
1.25
1.50
1.75
2.00
2.25
2.50
1.385
1.330
1.288
1.256
1.230
1.209
1.192
1.177
1.353
1.303
1.265
1.236
1.212
1.193
1.177
1.163
1.331
1.284
1.249
1.222
1.200
1.182
1.166
1.154
1.301
1.259
1.227
1.202
1.182
1.166
1.152
1.140
1.279
1.241
1.211
1.188
1.170
1.155
1.142
1.131
1.263
1.227
1.199
1.178
1.160
1.146
1.134
1.124
1.378
1.324
1.283
1.251
1.226
1.206
1.188
1.174
1.338
1.291
1.255
1.226
1.204
1.185
1.170
1.157
1.311
1.268
1.235
1.209
1.188
1.171
1.157
1.145
1.274
1.236
1.207
1.185
1.167
1.152
1.139
1.129
1.248
1.214
1.188
1.168
1.151
1.138
1.127
1.117
1.228
1.197
1.173
1.155
1.140
1.127
1.117
1.108
1.373
1.319
1.279
1.248
1.223
1.203
1.186
1.172
1.326
1.280
1.245
1.218
1.197
1.179
1.164
1.151
1.294
1.253
1.222
1.197
1.178
1.162
1.149
1.137
1.249
1.215
1.189
1.169
1.152
1.139
1.127
1.118
1.219
1.189
1.166
1.149
1.134
1.122
1.112
1.104
1.196
1.169
1.149
1.133
1.121
1.110
1.101
1.094
1.369
1.316
1.277
1.246
1.221
1.201
1.184
1.170
1.314
1.270
1.237
1.211
1.190
1.173
1.159
1.147
1.277
1.239
1.210
1.187
1.168
1.153
1.141
1.130
1.226
1.196
1.172
1.154
1.139
1.126
1.116
1.107
1.192
1.166
1.146
1.131
1.118
1.108
1.099
1.092
1.165
1.143
1.126
1.113
1.102
1.093
1.086
1.080
1.366
1.314
1.274
1.244
1.219
1.199
1.183
1.169
1.304
1.261
1.229
1.204
1.184
1.167
1.154
1.142
1.262
1.226
1.198
1.177
1.159
1.145
1.133
1.123
1.205
1.177
1.156
1.139
1.126
1.115
1.105
1.098
1.166
1.144
1.127
1.113
1.102
1.094
1.086
1.080
1.136
1.118
1.105
1.094
1.085
1.077
1.071
1.066
1.363
1.312
1.273
1.242
1.218
1.198
1.182
1.168
1.294
1.253
1.222
1.198
1.178
1.162
1.149
1.138
1.247
1.213
1.187
1.167
1.151
1.137
1.126
1.117
1.184
1.159
1.140
1.125
1.113
1.103
1.095
1.088
1.141
1.122
1.108
1.097
1.087
1.080
1.074
1.068
1.109
1.094
1.083
1.075
1.068
1.062
1.057
1.053
1.361
1.310
1.271
1.241
1.217
1.197
1.181
1.167
1.285
1.245
1.215
1.192
1.173
1.157
1.144
1.133
1.233
1.201
1.177
1.158
1.143
1.130
1.119
1.110
1.164
1.142
1.125
1.112
1.101
1.092
1.085
1.079
1.117
1.101
1.090
1.080
1.073
1.067
1.061
1.057
1.082
1.071
1.063
1.057
1.051
1.047
1.043
1.040
1.366
1.314
1.275
1.244
1.220
1.200
1.183
1.169
1.335
1.288
1.252
1.224
1.202
1.184
1.168
1.155
1.313
1.269
1.236
1.210
1.189
1.172
1.158
1.146
1.283
1.244
1.214
1.191
1.172
1.157
1.144
1.133
1.262
1.226
1.199
1.177
1.160
1.146
1.134
1.124
1.246
1.213
1.187
1.167
1.150
1.137
1.126
1.116
1.360
1.308
1.270
1.240
1.216
1.196
1.180
1.166
1.321
1.276
1.242
1.215
1.194
1.176
1.162
1.149
1.294
1.253
1.222
1.198
1.178
1.162
1.149
1.137
1.257
1.222
1.195
1.174
1.157
1.143
1.131
1.121
1.231
1.200
1.176
1.157
1.142
1.129
1.119
1.110
1.212
1.183
1.161
1.144
1.130
1.119
1.109
1.101
1.355
1.304
1.266
1.237
1.213
1.194
1.178
1.164
1.308
1.265
1.232
1.207
1.186
1.170
1.156
1.144
1.276
1.238
1.209
1.186
1.168
1.153
1.140
1.130
1.233
1.201
1.177
1.158
1.142
1.130
1.119
1.110
1.203
1.175
1.154
1.138
1.125
1.114
1.105
1.097
1.180
1.156
1.137
1.123
1.111
1.101
1.093
1.086
1.351
1.301
1.263
1.234
1.211
1.192
1.176
1.162
1.297
1.256
1.224
1.200
1.180
1.164
1.150
1.139
1.260
1.224
1.197
1.176
1.159
1.144
1.133
1.123
1.210
1.182
1.160
1.143
1.129
1.118
1.108
1.100
1.176
1.152
1.134
1.120
1.109
1.099
1.091
1.084
1.150
1.130
1.115
1.103
1.093
1.085
1.078
1.072
1.348
1.298
1.261
1.232
1.209
1.190
1.174
1.161
1.287
1.247
1.217
1.193
1.174
1.158
1.145
1.134
1.245
1.211
1.186
1.166
1.150
1.136
1.125
1.116
1.189
1.163
1.144
1.129
1.116
1.106
1.098
1.090
1.150
1.130
1.115
1.103
1.093
1.085
1.078
1.072
1.121
1.105
1.093
1.083
1.076
1.069
1.064
1.059
1.345
1.296
1.259
1.231
1.208
1.189
1.173
1.160
1.277
1.239
1.209
1.187
1.168
1.153
1.141
1.130
1.231
1.199
1.175
1.156
1.141
1.129
1.118
1.109
1.168
1.146
1.128
1.115
1.104
1.095
1.087
1.081
1.126
1.109
1.096
1.086
1.078
1.071
1.066
1.061
1.094
1.082
1.072
1.065
1.059
1.054
1.049
1.046
1.343
1.295
1.258
1.229
1.207
1.188
1.172
1.159
1.268
1.231
1.203
1.181
1.163
1.148
1.136
1.126
1.217
1.187
1.165
1.147
1.133
1.121
1.111
1.103
1.148
1.129
1.113
1.102
1.092
1.084
1.077
1.072
1.102
1.089
1.078
1.070
1.064
1.058
1.054
1.050
1.067
1.059
1.052
1.047
1.042
1.039
1.036
1.033
1.357
1.306
1.268
1.238
1.214
1.195
1.179
1.165
1.326
1.280
1.245
1.218
1.197
1.179
1.164
1.151
1.304
1.262
1.229
1.204
1.184
1.168
1.154
1.142
1.274
1.236
1.208
1.185
1.167
1.152
1.139
1.129
1.254
1.219
1.192
1.171
1.155
1.141
1.129
1.120
1.238
1.205
1.180
1.161
1.145
1.132
1.122
1.113
1.350
1.300
1.263
1.234
1.210
1.191
1.175
1.162
1.312
1.268
1.235
1.209
1.188
1.171
1.157
1.145
1.285
1.245
1.215
1.192
1.173
1.157
1.144
1.134
1.248
1.214
1.188
1.168
1.151
1.138
1.127
1.117
1.223
1.192
1.169
1.151
1.137
1.124
1.114
1.106
1.203
1.176
1.155
1.138
1.125
1.114
1.105
1.097
1.345
1.296
1.259
1.231
1.208
1.189
1.173
1.160
1.299
1.257
1.226
1.201
1.181
1.165
1.151
1.140
1.267
1.230
1.202
1.180
1.163
1.148
1.136
1.126
1.224
1.194
1.170
1.152
1.137
1.125
1.115
1.106
1.194
1.168
1.148
1.132
1.120
1.109
1.100
1.093
1.172
1.149
1.131
1.117
1.106
1.097
1.089
1.082
1.341
1.293
1.257
1.228
1.205
1.187
1.171
1.158
1.288
1.248
1.218
1.194
1.175
1.159
1.146
1.135
1.251
1.217
1.191
1.170
1.153
1.140
1.128
1.119
1.202
1.174
1.154
1.137
1.124
1.113
1.104
1.096
1.168
1.145
1.128
1.115
1.104
1.095
1.087
1.081
1.142
1.123
1.109
1.097
1.088
1.080
1.074
1.069
1.338
1.290
1.254
1.226
1.204
1.185
1.170
1.157
1.278
1.239
1.210
1.187
1.169
1.154
1.141
1.130
1.236
1.204
1.179
1.160
1.145
1.132
1.121
1.112
1.180
1.156
1.138
1.123
1.111
1.102
1.093
1.086
1.142
1.123
1.109
1.098
1.088
1.081
1.074
1.069
1.113
1.099
1.087
1.078
1.071
1.065
1.060
1.055
1.336
1.288
1.253
1.225
1.202
1.184
1.169
1.156
1.268
1.231
1.203
1.181
1.163
1.149
1.136
1.126
1.222
1.192
1.169
1.151
1.136
1.124
1.114
1.105
1.160
1.139
1.122
1.109
1.099
1.090
1.083
1.077
1.118
1.102
1.090
1.081
1.073
1.067
1.062
1.057
1.086
1.075
1.066
1.060
1.054
1.049
1.046
1.042
1.334
1.287
1.251
1.223
1.201
1.183
1.168
1.155
1.259
1.223
1.196
1.175
1.158
1.144
1.132
1.122
1.208
1.180
1.158
1.142
1.128
1.117
1.107
1.099
1.140
1.122
1.107
1.096
1.087
1.080
1.073
1.068
1.094
1.082
1.073
1.065
1.059
1.054
1.050
1.046
1.060
1.052
1.046
1.042
1.038
1.035
1.032
1.030
Table 6. B-Basis LEFs
1
n
N
1
1.5
27
2
3
4
5
L
R
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
17.5
20.0
22.5
25.0
27.5
30.0
32.5
35.0
2
3
1.00
1.25
1.50
1.75
2.00
2.25
2.50
1.00
1.25
1.50
1.75
2.00
2.25
2.50
1.00
1.25
1.50
1.75
2.00
2.25
2.50
1.211
1.182
1.160
1.143
1.129
1.118
1.108
1.100
1.183
1.158
1.140
1.125
1.113
1.103
1.095
1.088
1.164
1.142
1.125
1.112
1.101
1.093
1.085
1.079
1.137
1.119
1.105
1.094
1.085
1.078
1.072
1.066
1.119
1.103
1.091
1.082
1.074
1.068
1.062
1.058
1.104
1.091
1.080
1.072
1.065
1.060
1.055
1.051
1.205
1.177
1.156
1.139
1.126
1.115
1.105
1.098
1.170
1.148
1.130
1.116
1.105
1.096
1.088
1.082
1.146
1.127
1.112
1.100
1.091
1.083
1.076
1.071
1.114
1.099
1.087
1.078
1.071
1.065
1.060
1.055
1.091
1.079
1.070
1.063
1.057
1.052
1.048
1.045
1.074
1.064
1.057
1.051
1.046
1.042
1.039
1.036
1.200
1.173
1.153
1.136
1.123
1.112
1.103
1.096
1.159
1.138
1.122
1.109
1.099
1.090
1.083
1.077
1.131
1.114
1.100
1.090
1.081
1.074
1.069
1.063
1.092
1.080
1.071
1.064
1.058
1.053
1.049
1.045
1.066
1.057
1.051
1.046
1.041
1.038
1.035
1.032
1.046
1.040
1.035
1.032
1.029
1.026
1.024
1.023
1.197
1.170
1.150
1.134
1.121
1.111
1.102
1.094
1.149
1.129
1.114
1.102
1.093
1.085
1.078
1.072
1.117
1.101
1.090
1.080
1.073
1.067
1.061
1.057
1.072
1.063
1.056
1.050
1.045
1.042
1.038
1.036
1.042
1.037
1.032
1.029
1.026
1.024
1.022
1.021
1.019
1.017
1.015
1.013
1.012
1.011
1.010
1.009
1.194
1.168
1.148
1.132
1.120
1.109
1.100
1.093
1.140
1.122
1.107
1.096
1.087
1.080
1.073
1.068
1.103
1.090
1.079
1.071
1.065
1.059
1.054
1.050
1.053
1.046
1.041
1.037
1.034
1.031
1.028
1.026
1.019
1.017
1.015
1.013
1.012
1.011
1.010
1.010
1.192
1.166
1.146
1.131
1.118
1.108
1.099
1.092
1.132
1.114
1.101
1.090
1.082
1.075
1.069
1.064
1.090
1.079
1.070
1.062
1.057
1.052
1.048
1.044
1.035
1.031
1.027
1.024
1.022
1.020
1.019
1.017
1.190
1.165
1.145
1.130
1.117
1.107
1.098
1.091
1.123
1.107
1.095
1.085
1.077
1.070
1.065
1.060
1.078
1.068
1.060
1.054
1.049
1.045
1.041
1.038
1.017
1.015
1.014
1.012
1.011
1.010
1.009
1.009
1.195
1.168
1.148
1.133
1.120
1.109
1.101
1.093
1.167
1.145
1.128
1.114
1.104
1.094
1.087
1.080
1.148
1.129
1.114
1.102
1.092
1.084
1.077
1.072
1.122
1.106
1.094
1.084
1.076
1.069
1.064
1.059
1.104
1.090
1.080
1.072
1.065
1.059
1.055
1.051
1.090
1.078
1.069
1.062
1.056
1.051
1.047
1.044
1.189
1.163
1.144
1.129
1.116
1.106
1.098
1.090
1.155
1.134
1.118
1.106
1.096
1.088
1.081
1.075
1.131
1.114
1.101
1.090
1.082
1.075
1.069
1.064
1.099
1.086
1.076
1.068
1.062
1.057
1.052
1.048
1.077
1.067
1.059
1.053
1.048
1.044
1.041
1.038
1.060
1.052
1.046
1.041
1.038
1.034
1.032
1.029
1.184
1.160
1.141
1.126
1.114
1.104
1.095
1.088
1.144
1.125
1.110
1.099
1.089
1.082
1.075
1.070
1.116
1.101
1.089
1.080
1.072
1.066
1.061
1.056
1.078
1.068
1.060
1.054
1.049
1.045
1.041
1.038
1.052
1.045
1.040
1.036
1.033
1.030
1.027
1.025
1.032
1.028
1.025
1.022
1.020
1.018
1.017
1.016
1.181
1.157
1.138
1.124
1.112
1.102
1.094
1.087
1.134
1.116
1.103
1.092
1.083
1.076
1.070
1.065
1.102
1.089
1.078
1.070
1.064
1.058
1.054
1.050
1.058
1.051
1.045
1.040
1.037
1.034
1.031
1.029
1.028
1.025
1.022
1.020
1.018
1.016
1.015
1.014
1.005
1.005
1.004
1.004
1.003
1.003
1.003
1.003
1.178
1.154
1.136
1.122
1.110
1.100
1.092
1.086
1.125
1.109
1.096
1.086
1.078
1.071
1.065
1.061
1.089
1.077
1.068
1.061
1.056
1.051
1.047
1.043
1.039
1.034
1.030
1.027
1.025
1.023
1.021
1.019
1.006
1.005
1.004
1.004
1.004
1.003
1.003
1.003
1.176
1.153
1.135
1.120
1.109
1.099
1.091
1.085
1.117
1.101
1.090
1.080
1.073
1.066
1.061
1.057
1.076
1.066
1.059
1.053
1.048
1.044
1.040
1.037
1.021
1.019
1.017
1.015
1.014
1.012
1.011
1.011
1.175
1.151
1.133
1.119
1.108
1.098
1.090
1.084
1.108
1.094
1.083
1.075
1.068
1.062
1.057
1.053
1.064
1.056
1.049
1.044
1.040
1.037
1.034
1.031
1.004
1.003
1.003
1.003
1.002
1.002
1.002
1.002
1.186
1.161
1.142
1.127
1.115
1.105
1.096
1.089
1.159
1.138
1.122
1.109
1.099
1.090
1.083
1.077
1.140
1.122
1.108
1.096
1.087
1.080
1.073
1.068
1.114
1.099
1.088
1.079
1.071
1.065
1.060
1.056
1.096
1.084
1.074
1.066
1.060
1.055
1.051
1.047
1.082
1.072
1.063
1.057
1.052
1.047
1.043
1.040
1.180
1.156
1.138
1.123
1.111
1.102
1.093
1.086
1.147
1.127
1.112
1.101
1.091
1.083
1.077
1.071
1.123
1.107
1.095
1.085
1.077
1.070
1.065
1.060
1.091
1.079
1.070
1.063
1.057
1.052
1.048
1.045
1.069
1.060
1.053
1.048
1.043
1.040
1.037
1.034
1.052
1.046
1.040
1.036
1.033
1.030
1.028
1.026
1.176
1.152
1.134
1.120
1.109
1.099
1.091
1.084
1.136
1.118
1.104
1.093
1.084
1.077
1.071
1.066
1.108
1.094
1.083
1.075
1.068
1.062
1.057
1.053
1.070
1.061
1.054
1.049
1.044
1.040
1.037
1.035
1.044
1.039
1.034
1.031
1.028
1.026
1.024
1.022
1.025
1.021
1.019
1.017
1.016
1.014
1.013
1.012
1.173
1.150
1.132
1.118
1.107
1.097
1.090
1.083
1.126
1.110
1.097
1.087
1.079
1.072
1.066
1.061
1.094
1.082
1.073
1.065
1.059
1.054
1.050
1.046
1.051
1.044
1.039
1.035
1.032
1.029
1.027
1.025
1.021
1.018
1.016
1.015
1.013
1.012
1.011
1.010
1.170
1.147
1.130
1.116
1.105
1.096
1.088
1.082
1.117
1.102
1.090
1.081
1.073
1.067
1.061
1.057
1.081
1.071
1.062
1.056
1.051
1.047
1.043
1.040
1.032
1.028
1.025
1.022
1.020
1.019
1.017
1.016
1.168
1.146
1.128
1.115
1.104
1.095
1.087
1.081
1.109
1.095
1.084
1.075
1.068
1.062
1.057
1.053
1.068
1.060
1.053
1.047
1.043
1.039
1.036
1.034
1.014
1.012
1.011
1.010
1.009
1.008
1.008
1.007
1.166
1.144
1.127
1.114
1.103
1.094
1.086
1.080
1.101
1.088
1.077
1.069
1.063
1.058
1.053
1.049
1.056
1.049
1.044
1.039
1.036
1.033
1.030
1.028
3.4 COMBINED LOAD-LIFE APPROACH.
The life-factor approach requires an excessive test duration and, by itself, may not be practical
for full-scale test demonstration. In contrast, the LEF approach requires increasing the fatigue
loads so that the same level of reliability can be achieved with one test duration. However, for
hybrid structures, overloads may cause crack growth retardation, buckling, and premature failure
of some of the metal components. Another approach, which has been applied in the past, is a
combined approach using load and life factors, which is the general form of LEF given in
equation 9. The procedure in this approach would be to apply a combined life factor with the
load factor to achieve a compromise in the full-scale test requirements as well as the load
spectrum. This approach allows using a lower LEF as a trade-off for more life cycles, which
reduces the severity of the overload on metallic parts.
By combining equations 8 and 9, the LEF is defined in terms of test duration, as shown in
equation 12. This is a condensed form of equation 9.
L
 N R
LEF   F 
 N 
(12)
As shown in figure 15, LEF is significantly influenced by MSSP and MLSP. Thus,
improvements in these shape parameters for newer forms of materials that exhibit less scatter can
significantly reduce LEF.
1.70
1.60
LEF
1.50
L
1.40
1.30
n
0.50
1
1.25
1
2.50
1
2.50
5
1.20
1.10
1.00
0
10
20
30
40
50
60
70
80
90
100
R
Figure 15. Influence of Strength and Life Parameter on LEF
As shown in figure 16, a significant reduction in LEF is achieved simply by increasing the test
duration to 1.5 DLT. Furthermore, the influence of strength and life shape parameter on LEF
changes as the test duration is increased. For example, as the test duration is increased, the
28
influence of the fatigue-life shape parameter on LEF increases. This is understood, as NF is only
influenced by MLSP. Also, note for small test durations, the influence of MSSP on LEF is
significant due to the increased influence of load factor.
Figure 16. Influence of MSSP and MLSP on LEF (Combined Load-Life Approach)
Application of the combined load-life approach is illustrated in figure 17. The LEF is calculated
as the ratio of the maximum applied load to the design maximum fatigue stress. This curve is
generated by calculating LEF for several different test durations. Note that the LEF required for
the test duration (which is equal to the life factor, NF) is one, and NF is obtained from figure 14
for the corresponding fatigue-life shape parameter. If the design maximum load in the repeated
load test (PF) is increased to the mean residual strength at one lifetime (PT), then the A- or Bbasis residual strength of the structure would be equivalent to the design maximum fatigue stress.
Thus, a successful repeated load test to one lifetime at applied stress, PT, or a repeated test to NF
at applied stress, PF, (no LEF) would both demonstrate the corresponding reliability.
Furthermore, a successful repeated load test to a test duration (less than NF) with the
corresponding LEF would demonstrate the same level or reliability on the design lifetime.
29
  L  1 
  L 

PM – Static Strength
NF
Combined Approach
LEF (N=1)
1
 ln ( p) 


2
    ( 2 n )  


  2 n  
L
PF – Design
Maximum
Fatigue Stress
1
10
TEST DURATION (N)
100
Figure 17. Combined Load-Life Approach for Composite Structures
Figure 18 shows the A- and B-basis LEF requirements based on the number of test specimens
using equation 9 for the shape parameters reported in reference 2, R = 20 and L = 1.25. It
shows that the increased number of test specimens reduces the LEF requirements. However, in
large-scale testing, only a single fatigue test is conducted due to high cost and long test duration.
This figure also shows that the A-basis requirements on LEF are significantly higher than for
B-basis.
30
1.50
Strength Shape Parameter:  R = 20
Life Shape Parameter:  L = 1.25
1.40
LEF
1.30
1.20
A-basis
n=1,2,5,10,15, and 30
1.10
B-basis
n=1,2,5,10,15, and 30
1.00
0
5
10
15
20
Test Duration, N
1 DLT
1.5 DLT
2 DLT
Sample
Size
A-Basis B-Basis A-Basis B-Basis A-Basis B-Basis
1
1.324
1.177
1.291
1.148
1.268
1.127
2
1.308
1.163
1.276
1.134
1.253
1.114
5
1.291
1.148
1.259
1.120
1.237
1.100
10
1.282
1.140
1.250
1.111
1.227
1.091
15
1.277
1.135
1.245
1.107
1.223
1.087
30
1.270
1.130
1.239
1.101
1.217
1.082
Figure 18. A- and B-Basis LEF Requirements With Respect to the Number of Test Specimens
3.4.1 Application of LEF to a Load Spectrum.
The combined load-life approach can be used in two different ways: (1) to apply the same LEF,
which is calculated for a certain test duration, to the entire spectrum or (2) to apply a different
LEF for different load blocks in the spectra based on the severity of enhanced load, i.e., cycles
that have high loads are repeated for a longer test duration (with lower LEF) than the rest of the
spectrum. This approach is particularly useful for hybrid structures that exhibit metallic
component failure due to high LEF (overloads) and to avoid premature failure due to buckling.
Both of these applications are illustrated in figure 19.
31
Original Spectrum
Enhanced Spectrum
N = N1 < NF
LEF(N1) > 1
N = N1 < N2
LEF(N1) > LEF(N2)
(a) Combined Load-Life Test
N = N2 [or NF]
LEF = LEF(N2) [or 1]
(b) Combined Load-Life Spectrum
Figure 19. Application of Combined Load-Life Approach
One common practice for composite full-scale test substantiation is a 2 DLT, which is adopted
from metallic structural certification (figure 14) using the design spectrum with the
corresponding LEF under room temperature ambient (RTA) test conditions. Often, the test
duration of 1.5 DLT is used with the corresponding LEF. The LEF approach accounts for the
variability in design details and loading modes. To account for the service environmental effects
on composites, additional factors are calculated from the design allowable tests. These
additional factors for composite structures account for the difference between composite and
metallic structure during design and analysis and are beyond what is normally done for metallic
certifications. Such factors, accounting for both moisture and temperature effects on composites,
depend significantly on the material system and the lay-up configuration and can be as high
as 1.4.
In metallic structures, severe flight loads result in crack growth retardation. Therefore, it is
common practice to clip the peak loads (i.e., reduce the peak loads to the clipping level, without
omitting loads) after careful consideration of the appropriate clipping levels. In contrast, such
severe flight loads significantly contribute to flaw growth in composite structures and reduce the
fatigue life. Therefore, fatigue spectrum clipping must not be done during structural testing of
composite structures. When additional LEFs and environmental factors are applied to composite
load spectrum, the cumulative effects on test loads can be significantly higher than the clipping
levels for metallic components. For hybrid structures, this can be addressed through the
combined load-life spectrum approach shown in figure 19(b). Alternatively, fatigue tests with
32
LEFs (not combined with environmental compensation factors) can be performed by including
periodic static tests with the environmental factors applied to relieve the severity of high loads.
To reduce the test duration, the fatigue test spectrum is truncated by eliminating the segments
with stress levels below an endurance limit (stress level corresponds to an infinite life). The
endurance limit of a particular composite material varies, based on parameters such as the lay-up
configuration, test environment, and stress ratio. The S/N curves that are generated to obtain the
life shape parameters can be used to determine the endurance limit for different design details of
a composite structure. The life factor and the LEF are to be applied after truncation of the load
spectrum as shown in figure 20.
Figure 20. Fatigue Test Spectrum Development for Composite Structural Test
Composite material properties are susceptible to temperature and moisture. Therefore, when a
full-sale test is conducted at RTA conditions, environmental compensation factors are applied to
the load spectrum. The environmental factors are recommended to be applied to the truncated
load spectrum. Although this approach provides an efficient way to ensure structural life
reliability, the cumulative effects of the above-mentioned enhancement factors may result in an
undesirably high cumulative LEF. For these cases, an approach similar to that shown in
figure 19(b) is recommended to reduce the required LEFs for high-spectrum loads, i.e., test for
33
additional life. However, these additional high-stress cycles must be spread throughout the
spectrum so that the damage growth mechanism is not adversely altered, and the practical limits
of spectrum load sequence must be preserved. Although there are no significant load-sequencing
effects on fatigue life, composites are extremely sensitive to variation in the number of high
loads in the fatigue spectrum [2]. It is imperative that the effects of these parameters do not
change the fatigue failure mode (and the failure mechanism) or reach the static strength of the
structure. Furthermore, the application of load enhancements must preserve the stress ratio of
each load cycle throughout the spectrum so that the fatigue damage mechanism and the life are
not artificially influenced.
The LEF can be applied to the fatigue spectrum in several ways: (1) to 1-g mean fatigue load,
(2) to amplitude, and (3) to minimum/maximum load. These three approaches are shown in
equations 13 through 15 for a typical aircraft loads application.

 Load
Pmean   Load 1 g   LEF  
 g



  g 




 Load 
Pamplitude  Load1 g   
  g  LEF 
 g 




 Load 
PMin / Max  Load 1 g   
  g   LEF
 g 


(13)
(14)
(15)
where
(Load1-g) = mean fatigue load (i.e., 1-g maneuver shear/moment/torque) of a load block

g
= amplitude with respect to 1-g fatigue load
 Load 

 = load (shear/moment/torque) per g
 g 
When applying the LEFs to mean fatigue loads, as shown in equation 13, the mean load is offset
in either the positive (for positive mean loads) or negative (for negative mean loads) direction.
For cycles with load reversal (stress ratio R <0), this causes a reduction in load magnitudes in the
opposite loading direction, i.e., shifts the mean load, as shown in figure 21. Consequently, this
alters the damage growth caused by reversible loads to the composite structure. Furthermore, for
higher LEF values, this may convert a tension-compression cycle to a tension-tension cycle or
compression-compression cycle for positive and negative enhancement of mean loads,
respectively. Specimen-level data for composite materials show that reversible load cases (R <0)
are critical and have a significantly lower fatigue life than that of tension-tension or
compression-compression (R >0) cases. Similarly, application of LEF to stress amplitude about
the mean load, as shown in equation 14, results in an alteration of the stress ratio and possibly
load reversal for high LEFs. Therefore, equations 13 and 14 are not recommended for applying
34
the LEF to a spectrum loading with negative stress ratios (tension-compression loading) to avoid
changes to stress ratios and unintentional reduction in fatigue damage to the test article.
1
2
Moment (in-lbs)
1
0
00
8
0
6
0
Enahnced Mean
LEF
4
0
2
0
1-g Mean
(+)
0
0
(-)
-2
0
-4
0
Reduction in reversed load
Figure 21. Application of LEF Only to Mean Load
Application of the LEF to both 1-g mean and amplitude, as in equation 15, results in
considerably high loads but maintains the same stress ratios throughout the spectrum. Therefore,
full-scale fatigue test spectrum loads are generated by applying the LEF to the minimum and
maximum shear-moment torque (SMT) loads so that the reversible loads are not shifted but
rather enhanced, depending on the sign of the maximum or minimum SMT load, and the stress
ratio is maintained after load enhancement.
3.4.2 Generating Life Factor and LEFs.
To generate reliable LEFs with a statistical significance, a large number of test data is required.
Figure 22 shows the minimum required test matrix for generating strength and life shape
parameters so that life factors and LEFs for a particular structure can be developed. Such a test
matrix must consider the following:

Critical design details, loading modes, and stress ratios must be represented in coupon
and element level.

Specimens must be fabricated from a minimum of three distinctive material batches,
unless otherwise proven that significant batch variability does not exist through laminalevel statistics (i.e., alpha = 0.01 significance level using the Anderson-Darling test [44]).

Excluding the static stress level, a minimum of three fatigue stress levels must be
included in the test matrix for each S/N curve. At least two stress levels must show
fatigue failures for analysis techniques that utilize residual strength of runouts, or else all
stress levels must have a minimum of six fatigue failures.
35

Based on the fatigue analysis technique (i.e., analysis of individual stress levels or
pooling fatigue data across stress levels using either joint Weibull or Sendeckyj analysis),
the minimum recommended number of specimens, as shown in figure 22, must be
included.

A minimum of six data sets for static are included (at least six Weibull shape parameters)
in the static-strength scatter analysis for generating the strength shape parameter.

A minimum of six data sets for fatigue are included (at least six Weibull shape
parameters) in the fatigue-life scatter analysis for generating the life shape parameter.
Fatigue-Critical Design Details - Cyclic
Test R ratio (3 Stress Levels)
R1
R2
R3
R4
Design Detail
Test Method
Loading
Condition
Environmental
Condition
Static-Critical
Design Details
1
Method 1
1
1
Bx6
2
Method 2
2
1
Bx6
3
Method 3
3
1
Bx6
4
Method 4
4
1
Bx6
5
Method 5
5
1
Bx6
5
Method 5
5
2
Bx6
1
Method 1
1
1
Bx3xF
2
Method 2
2
1
Bx3xF
3
Method 3
3
1
4
Method 4
4
1
5
Method 5
5
1
Bx3xF
5
Method 5
5
2
Bx3xF
NOTES: Static - B x (# of test specimens); B =
Fatigue - B x (# of stress levels) x F; F =
Bx3xF
Bx3xF
1, if no significant batch variability exists in lamina level
3, if significant batch variability exists in lamina level
2 for pooled fatigue analysis techniques
6 for individual fatigue analysis techniques
Figure 22. Minimum Test Requirements for Generating Life Factors and LEFs
The composite data analyzed in reference 4 suggest that the LEFs shown in figure 18 are
conservative for modern composites as a result of the improvements in materials and process
techniques and test methods (i.e., less scatter in test data). Therefore, in the absence of sufficient
test data, the values in figure 18 can be used during large-scale test substantiation. However,
some of the modern, complex composite material forms or any composite material that exhibits
large data scatter may have to be examined to ensure that the strength and life shape parameters
are at least equivalent or better than the values shown in figure 18. This can be achieved through
a minimum of a fatigue-life scatter analysis of representative coupon/element level S/N data
obtained for one fatigue-critical design detail and a representative stress ratio (i.e., open-hole test
with R=0), as shown in figure 23.
36
Figure 23. Minimum Requirements to be Fulfilled Prior to Using LEFs in Figure 18 for a
Composite Structural Test
3.5 SCATTER ANALYSIS COMPUTER CODE.
The scatter analysis conducted for this report was carried out using a combination of Microsoft®
Excel®, Visual Basic®, and ReliaSoft Weibull software [42]. Sendeckyj analysis was coded in
Microsoft .NET Framework software. To alleviate the dependency on multiple software
packages, the complete analysis with multiple analysis options was coded using a Microsoft
Visual Basic macro that could be run in Microsoft Excel.
A user-friendly computer code, Scatter Analysis Computer Code (SACC) (figure 24), was
developed so that the fatigue test data could be analyzed using the individual Weibull
distribution, joint Weibull distribution, and Sendeckyj equivalent static-strength model.
Appendix C contains the flow diagram for SACC. This program was designed to guide the
analyst to select the most suitable approach for a given set of test data.
37
Figure 24. Scatter Analysis Using SACC
4. STATIC STRENGTH DATA SCATTER ANALYSIS.
An extensive material database is available in reference 2. Using this data, static scatter analysis
was conducted to support U.S. Navy F/A-18 certification. These data represented several
structural details and variables such as laminate lay-up, loading mode, load transfer, specimen
geometry, and environment. To improve the accuracy of the Weibull analysis, only the data sets
38
containing six or more specimens were included. Also, the U.S. Navy F/A-18 certification
program only included autoclaved 350°F-cure graphite-epoxy materials. This analysis was
conducted primarily on fiber-dominated failures. In addition, these data were summarized
primarily for laminated construction and did not include sandwich construction or bonded joints.
The goal of the current research was to produce data for materials commonly used in aircraft
applications and to promote the development of this type of data, not only for individual
certification plans, but also for shared databases.
4.1 STRUCTURAL DETAILS FOR STATIC SCATTER ANALYSIS.
The current research obtained the static-strength data from several material databases, described
in section 2.1. These data included details such as bonded joints, sandwich details, impact
damage, disbonds, lightening strikes, process variability, in-plane shear and interlaminar shear
specimens in addition to the variables studied in the U.S. Navy F/A-18 certification program.
These structural details were included in the analysis to represent design variables or details in
present aircraft applications. Data generated from coupons and elements were used to
investigate the dependence of the static-strength shape parameter, which is a representation of
static data scatter, on various coupon geometries, loading modes, environments, and lay-ups.
The degree to which these parameters affect the overall LEF factors on a parametric basis is
discussed in section 5.4. Several examples are shown for obtaining MSSP, or R, by pooling
different data sets. When pooling data to estimate MSSP, the user is advised to select
appropriate design details that are applicable to a certain application.
The material databases described in this section represent typical examples of commonly used
composite materials. These materials are used in several ongoing certification programs and
certified general aviation aircraft. The databases include coupon-level, static-strength data for
the following primary variables:

Different lay-ups—hard, quasi-isotropic, and soft (typically 50/40/10, 25/50/25, and
10/80/10, respectively, for unidirectional material and 40/20/40, 25/50/25, and 10/80/10,
respectively, for fabric material) and all ±45° plies (0/100/0)

Environments—cold temperature dry (CTD), RTA, room temperature wet (RTW),
elevated temperature dry (ETD), elevated temperature wet (ETW)

Tension—unnotched, OH, filled hole (FH)

Compression—unnotched, OH, FH

Bearing—single shear, double shear, bearing-bypass

In-plane shear—VNRS

Interlaminar shear—double-notched compression, short-beam shear
39
In addition to the coupon-level data, element-level static-strength data were included in the
analysis, representing the following design details/requirements:




Sandwich—core materials, facesheet
Bonded joints—single-lap shear, picture frame
Damage tolerance—CAI, TAI, sandwich
Four-point bending—laminate, sandwich
The following sections include the generation of the shape parameter for several different
material systems accounting for the effects of different geometries, environments, lay-ups, and
loading modes.
4.1.1 Advanced Composites Group AS4/E7K8 3K Plain-Weave Fabric.
The static-strength test results of AS4/E7K8 plain-weave fabric (AS4-PW) for both RTA and
ETW environmental conditions are shown in table 7.
Table 7. Static-Strength Test Results for AS4-PW
Specimen Configuration
Lay-Up
Test Description
10/80/10
OHT
OHC
SLS–C (t=0.01″)
SLS–T (t=0.01″)
SLS–T (t=0.01″)
No antibuckling
SLS–T (t=0.06″)
DNC
CAI – BVID (20 plies)
CAI – VID (40 plies)
Sandwich
4PB – HRH 10
0/100/0
TAI – BVID
TAI – VID
OHC
25/50/25
OHC
Unimpacted
CAI – BVID
CAI – VID
CAI – LID
40/20/40
CAI – VID
STDEV = Standard deviation
CV = Coefficient of variation
BVID = Barely visible impact damage
RTA Strength
(ksi)
Average
STDEV
43.455
0.773
40.472
1.571
5.532
0.381
4.528
0.163
2.301
0.058
5.057
3.988
34.605
30.263
0.145
21.875
15.118
17.799
45.375
55.736
36.025
29.671
25.445
31.845
0.591
0.160
1.541
0.814
0.003
0.350
0.626
0.387
1.624
0.839
0.851
0.891
0.692
1.026
CV
1.778
3.882
6.880
3.601
2.532
11.685
4.022
4.454
2.690
2.071
1.600
4.143
2.172
3.579
1.505
2.361
3.003
2.721
3.223
ETW Strength
(ksi)
Average
STDEV
CV
35.175
0.296
0.842
28.579
1.156
4.045
2.038
3.004
0.330
0.154
16.196
5.130
0.128
11.912
0.003
1.295
2.388
10.876
32.019
1.347
4.208
VID = Visible impact damage
LID = Large impact damage
40
Using ReliaSoft® Weibull software, the shape parameter ( α̂ ) and the scale parameter ( β̂ ) of each
data set were obtained and are shown in table 8.
Table 8. Weibull Parameters for Static-Strength Distributions of AS4-PW
Specimen Configuration
Lay-Up
10/80/10
Test Description
OHT
Test
Environment
OHC
SLS-C (t=0.01″)
SLS-T (t=0.01″)
SLS-T (t=0.01″)
No antibuckling
SLS-T (t=0.06″)
DNC
CAI-BVID (20 plies)
CAI-VID (40 plies)
7Sandwich 4PB-HRH 10
0/100/0
TAI-BVID
TAI-VID
OHC
25/50/25
OHC
40/20/40
CAI-BVID*
CAI-VID*
CAI-LID*
CAI-VID
Weibull Statistics
β̂ (ksi)
RTA
ETW
RTA
ETW
RTA
RTA
RTA
α̂ 
58.036
61.970
26.930
33.290
22.665
31.165
40.072
43.826
35.653
41.205
29.081
5.683
4.602
2.239
n
6
6
6
8
6
5
6
RTA
ETW
RTA
ETW
RTA
RTA
RTA
ETW
RTA
RTA
RTA
ETW
RTA
ETW
RTA
RTA
RTA
RTA
12.358
6.919
28.130
23.845
35.461
49.383
47.621
43.177
44.694
34.344
63.247
11.766
33.424
28.157
45.771
32.222
36.676
32.984
5.286
2.174
4.061
3.072
35.185
30.608
0.146
0.129
17.992
15.379
22.046
12.431
46.101
32.613
36.413
30.103
25.776
32.324
6
8
6
6
3*
6
6
6
5
6
6
5
6
6
6
6
6
6
* These shape parameters are not included in calculating MSSP. They were performed to investigate the data
scatter with respect to the damage intensity.
BVID = Barely visible impact damage
VID = Visible impact damage
LID = Large impact damage
41
The Weibull distribution of the shape parameters ( α̂ ) of RTA and ETW data shown in table 8
are compared against the pooled Weibull distribution of RTA and ETW data in table 9 (denoted
by All). Weibull statistics obtained for these three distributions of shape parameters from MLE
and both X and Y rank regression (RRX and RRY, respectively) are shown in table 9. Using the
shape and scale parameters,  and , respectively, the modal shape parameter corresponding to
the distribution of α̂ ’s, which is denoted as Modal, was calculated for each case. Unlike the cases
of test data distributions, the scale parameters here do not have units as they correspond to the
distributions of shape parameters ( α̂ ) obtained for different test data sets.
Table 9. Weibull Statistics for Combined Distribution of Scatter in Static-Strength
Distributions of AS4-PW
Analysis
Method
MLE
RRX
RRY
Weibull
Parameter
α
β
αModal
α
β
αModal
α
β
αModal
Analysis Cases
All
RTA
ETW
2.514 3.1323
1.782
39.387 41.777 33.629
32.193 36.950 21.183
2.188
2.900
1.441
39.882 41.980 34.452
30.167 36.284 15.148
2.103
2.790
1.417
40.285 42.288 34.665
29.644 36.068 14.621
The probability density function of Weibull shape parameters and the reliability plot for both
RTA and ETW static data (combined case denoted as All) are shown in figure 25. The shape
and scale parameters for this distribution from MLE are 2.514 and 39.387, respectively. The
corresponding modal value is 32.193. The difference between Weibull statistics obtained from
different analysis methods (i.e., MEL, RRX, and RRY) is least significant for the RTA data set
while it is most significant for the ETW data set. The scatter in ETW data sets is reflected in the
combined case for all three analysis methods. Statistics from MLE portray the least scatter for
all three cases while RRY portrays the most scatter. For large data sets, i.e., more than 20-30
samples, MLE produces reliable Weibull statistics, while for small data sets, RRX tends to
produce relatively accurate data. A procedure to obtain reliable Weibull statistics and
recommendations is documented in section 3.4.2. For example, if the structure does not contain
adhesive joints, the shape parameters for adhesive static-strength distributions do not have to be
pooled to determine MSSP.
42
Modal â
Mean â
(a)
90% Confidence
bounds (two sided)
on reliability
(b)
Figure 25. (a) Probability Density Function and (b) Reliability Plot of Shape Parameters
for AS4-PW Static Strength Distributions
43
4.1.2 Toray T700/#2510 Plain-Weave Fabric.
Toray T700/#2510 plain-weave fabric (T700-PW) data from several material databases (section
2.1) are analyzed in this section. Static strength results obtained from FAA-LEF data are shown
in table 10.
Table 10. Static-Strength Test Results for T700-PW
Specimen Configuration
Lay-Up
10/80/10
40/20/40
Sandwich
Test Description
OHT
OHC
SLS-T (t = 0.06″)
DNC
CAI-BVID
4PB-HRH 10
RTA Strength
(ksi)
Standard
Average
Deviation
41.686
0.889
34.986
0.923
5.064
0.197
3.007
0.197
43.408
0.610
0.137
0.003
CV
2.133
2.639
3.900
6.568
1.405
2.333
ETW Strength
(ksi)
Standard
Average
Deviation
38.961
0.886
28.626
0.788
CV
2.273
2.752
3.242
0.179
5.530
0.125
0.005
3.626
BVID = Barely visible impact damage
CV = Coefficient of variation
The corresponding Weibull statistics are shown in table 11 along with the number of specimens
used for analysis. Although the Weibull analysis was conducted for 40/20/40 CAI specimens,
the shape parameter was not included in the analysis for calculating MSSP of AS4-PW because
this data set had less than the minimum recommended number of specimens for a reliable
analysis.
Table 11. Weibull Parameters for Static-Strength Distributions of T700-PW
Specimen Configuration
Lay-Up
10/80/10
Test Description
OHT
OHC
SLS-T (t = 0.06″)
DNC
40/20/40
Sandwich
CAI
4PB-HRH 10
Weibull Statistics
Test Environment
α̂ 
β̂ (ksi)
RTA
ETW
RTA
ETW
RTA
RTA
ETW
RTA
RTA
ETW
48.872
45.242
41.540
40.170
36.927
17.989
25.000
67.713
42.068
42.190
42.108
39.387
35.408
28.989
5.144
3.094
3.317
43.700
0.139
0.127
*Not included in combined analysis due to insufficient data points.
44
n
6
6
6
7
6
6
6
3*
6
6
Forty-eight additional FAA-LVM data sets [33] containing 863 specimens for T700-PW were
added to the static-strength scatter analysis. These data sets represent hard (40/20/40, table 12),
quasi-isotropic (25/50/25, table 13), and soft (10/80/10, table 14) lay-up sequences, as well as a
wide range of loading modes and environmental conditions (CTD, RTA, and ETW).
Furthermore, each data set contains specimens from three distinct material batches.
Distribution of shape parameters calculated for T700-PW using the static data in FAA-LEF and
FAA-LVM is shown in figure 26. The scatter in shape parameters of 10/80/10 laminate is
significantly higher than for the other two laminate stacking sequences.
Table 12. Weibull Parameters for Static-Strength Distributions of
40/20/40 T700-PW (FAA-LVM)
Test Description
Single-shear bearing tension
Double-shear bearing tension
Bearing-bypass 50% compression
Bearing-bypass 50% tension
[t/D = 0.475]
Bearing-bypass 50% tension
[t/D = 0.570]
Bearing-bypass 50% tension
[t/D = 0.712]
Bearing-bypass 50% tension
[t/D = 0.949]
Unnotched tension
Unnotched compression
Open-hole compression
Filled-hole tension
Filled-hole tension
Filled-hole tension
V-notched rail shear
Open-hole tension [w/D = 3]
Open-hole tension [w/D = 4]
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 8]
Test
Environment
RTA
RTA
RTA
RTA
Shape
Parameter, α̂
45.409
49.696
42.102
40.040
Number of
Specimens
18
30
15
18
RTA
42.594
18
RTA
43.426
15
RTA
38.198
18
RTA
RTA
RTA
CTD
RTA
ETW
RTA
RTA
RTA
RTA
RTA
29.820
20.584
30.453
29.908
20.296
25.192
59.208
20.594
27.054
27.202
25.441
18
18
19
19
19
18
18
18
18
20
18
45
Table 13. Weibull Parameters for Static-Strength Distributions of
25/50/25 T700-PW (FAA-LVM)
Test Description
Double-shear bearing tension
Double-shear bearing tension
Double-shear bearing tension
Single-shear bearing tension
Single-shear bearing tension
Single-shear bearing tension
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Unnotched compression
Open-hole compression
Open-hole compression
Open-hole compression
V-notched rail shear
Test
Environment
CTD
RTA
ETW
CTD
RTA
ETW
RTA
RTA
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
RTA
Shape
Parameter, α̂
25.721
43.827
34.775
28.956
18.132
33.850
44.264
48.028
35.816
34.049
25.223
51.153
40.186
38.383
31.498
27.074
23.676
34.475
46.999
33.319
16.458
Number of
Specimens
18
18
18
18
18
18
15
15
18
18
21
18
18
18
18
18
19
19
18
21
18
Shape parameters obtained from static-strength distributions were combined for several different
analysis scenarios to investigate the degree to which these parameters affect the MSSP of
T700-PW (table 15). Adhesively bonded T700-PW element data from FAA-EOD material
database are included in section 4.1.9.
46
Table 14. Weibull Parameters for Static-Strength Distributions of
10/80/10 T700-PW (FAA-LVM)
Test Description
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched compression
Open-hole compression
V-notched rail shear
V-notched rail shear
V-notched rail shear
Shape
Parameter, α̂
65.445
74.360
51.713
58.084
36.056
50.909
9.963
17.278
13.103
Test
Environment
RTA
RTA
RTA
RTA
RTA
RTA
CTD
RTA
ETW
Number of
Specimens
15
15
8
19
18
18
18
19
18
80
Shape Parameter
60
40
20
0
0
10
40/20/40
20
3025/50/25
40
50
60
10/80/10
70
80
Sandwich
Datasets
Figure 26. Shape Parameters for T700-PW Static-Strength Distributions
47
Table 15. Summary of Weibull Shape Parameter Analysis of T700-PW
Weibull Statistics
Analysis Case
Analysis Variable
Database
Test environment
Lay-up
Loading mode
FAA-LVM
FAA-LEF
Combined
CTD
RTA
ETW
40/20/40
25/50/25
10/80/10
All bearing
OH/FH
VNRS
Unnotched

2.801
5.405
2.961
3.164
2.912
4.545
3.391
4.139
1.925
3.412
3.550
1.449
3.379

40.038
41.184
40.307
34.467
43.359
31.252
38.218
37.585
47.172
46.947
35.976
25.941
39.728
Modal
34.196
39.654
35.071
30.569
37.524
29.589
34.477
35.155
32.239
42.409
32.773
11.559
35.809
Number of
Data Sets
48
9
57
8
32
8
18
21
9
17
16
5
10
4.1.3 Toray 7781/#2510 8-Harness Satin-Weave Fabric.
Toray 7781/#2510 8-harness satin-weave fabric (7781-8HS) data from the FAA-LEF material
database are analyzed in this section. The static-strength test results for 7781-8HS are shown in
table 16, and the corresponding Weibull statistics are shown in table 17.
Table 16. Static-Strength Test Results for 7781-8HS
Specimen Configuration
Test
Lay-Up
Description
10/80/10 OHT
OHC
DNC
Sandwich 4PB-HRH 10
RTA Strength
(ksi)
Standard
Average Deviation
26.808
0.366
33.227
0.324
3.494
0.170
0.139
0.002
CV = Coefficient of variation
48
CV
1.364
0.974
4.861
1.795
ETW Strength
(ksi)
Standard
Average Deviation
CV
21.120
0.186
0.882
21.824
0.335
1.533
2.364
0.264
11.167
0.128
0.002
1.206
Table 17. Weibull Parameters for Static-Strength Distributions of 7781-8HS
Specimen Configuration
Lay-Up
10/80/10
Weibull Statistics
Test
Environment
Test
Description
OHT
RTA
ETW
RTA
ETW
RTA
ETW
RTA
ETW
OHC
DNC
Sandwich
4PB-HRH 10
α̂ 
76.115
116.288
104.824
62.351
32.221
10.116
80.260
86.690
β̂
(ksi)
26.983
21.211
33.385
21.998
3.561
2.476
0.142
0.128
n
6
6
6
6
6
6
6
6
Table 18 shows the analysis results for the Weibull distribution of shape parameters shown in
table 17. Compared to RRX and RRY analyses, MLE data indicate significant skewness,
possibly due to an insufficient number of data sets. It is important to explore the other two
regression techniques for such cases. For this case, the RRX method was selected to determine
the static-strength shape parameter.
Table 18. Weibull Statistics for Combined Distribution of 7781-8HS
Analysis
Method
MLE
RRX
RRY

Analysis Cases
All
RTA
2.185
3.295

79.512
82.038
Modal
60.092
73.510

1.438
2.079

83.239
84.814
Modal
36.416
61.868

1.278
1.868

87.043
87.042
Modal
26.385
57.748
Weibull
Parameter
49
4.1.4 Toray T700/#2510 Unidirectional Tape.
Toray T700/#2510 unidirectional tape (T700-UT) data from the FAA-LVM material database
[33] are analyzed in this section. This database contains 853 T700-UD specimens from 47 data
sets. These data sets represent hard (50/40/10), quasi-isotropic (25/50/25), and soft (10/80/10)
lay-up sequences, as well as a wide range of loading modes and environmental conditions (CTD,
RTA, and ETW). Furthermore, each data set contains specimens from three distinct material
batches. Shape parameters obtained from static-strength distributions were combined for several
different analysis scenarios to investigate the degree to which these parameters affect the MSSP
of T700-UT (table 19). The distribution of shape parameters calculated for T700-UT using the
static data in FAA-LVM is shown in figure 27. As for T700-PW, scatter in the distribution of
10/80/10 static-strength shape parameters is significantly higher than for the other two laminate
stacking sequences.
Table 19. Summary of Weibull Shape Parameter Analysis of T700-UT
Analysis
Case
Analysis
Variable
All
Test
environment
T700-UT
CTD
RTA
ETW
50/40/10
25/50/25
10/80/10
All bearing
OH/FH
V-notch rail shear
Unnotched
Lay-up
Loading
mode
Weibull Statistics

2.255
3.465
2.174
3.139
3.006
2.246
1.648
2.772
4.197
2.471
3.949
50

37.176
35.279
39.519
29.188
37.479
36.083
38.205
47.451
30.659
15.719
32.859
Modal
28.671
31.977
29.763
25.830
32.760
27.760
21.678
40.375
28.734
12.743
30.517
Number of
Data Sets
47
8
31
8
17
21
9
17
16
5
10
80
Shape Parameter
60
40
20
0
0
10
50/40/10
20
3025/50/2540
50
60
10/80/10
70
Datasets
Figure 27. Shape Parameters for T700-UT Static-Strength Distributions
4.1.5 Advanced Composites Group AS4C/MTM45 Unidirectional Tape.
ACG AS4C/MTM45 12K unidirectional tape (AS4C-UT) data from the FAA-LVM material
database [33] are analyzed in this section. This database contains 1151 AS4C-UT specimens
from 86 data sets. These data sets represent hard (50/40/10), quasi-isotropic (25/50/25), and soft
(10/80/10) lay-up sequences, as well as a wide range of loading modes and environmental
conditions (CTD, RTA, ETD, and ETW). Most of these data sets contain specimens from
multiple distinct material batches. Shape parameters obtained from static-strength distributions
were combined for several different analysis scenarios to investigate the degree to which these
parameters affect the MSSP of AS4C-UT (table 20). The distribution of shape parameters
calculated for AS4C-UT using the static data in FAA-LVM is shown in figure 28.
Table 20. Summary of Weibull Shape Parameter Analysis of AS4C-UT
Weibull Statistics
Analysis Case
All
Test environment
Analysis Variable
AS4C-UT
CTD
RTA
ETW

2.151
2.213
3.094
1.905
51

34.779
31.603
35.541
35.777
Modal
26.004
24.084
31.329
24.208
Number of
Data Sets
86
16
30
38
Table 20. Summary of Weibull Shape Parameter Analysis of AS4C-UT (Continued)
Weibull Statistics
Analysis Case
Analysis Variable
Lay-up
50/40/10
25/50/25
10/80/10
Lamina
All bearing
OH/FH
Unnotched
Loading mode

2.040
2.731
2.378
3.0435
2.729
2.632
2.252

36.953
36.635
46.883
25.5518
33.131
36.646
51.868
Number of
Data Sets
Modal
26.560
31.002
37.267
22.4171
28.028
30.561
39.964
14
19
19
34
6
30
12
105.8
80
Shape Parameter
60
40
20
50/0/50
0
0
10
50/40/10
2025/50/25
30
4010/80/10
50
60
70 Lamina
80
90
100
Datasets
Figure 28. Shape Parameters for AS4C-UT Static-Strength Distributions
4.1.6 Advanced Composites Group AS4C/MTM45 5-Harness Satin-Weave Fabric.
ACG AS4C/MTM45 five-harness satin-weave fabric (AS4C-5HS) data from the FAA-LVM
material database [33] are analyzed in this section. This database contains 1083 AS4C-5HS
specimens from 78 data sets. These data sets represent hard (40/20/40), quasi-isotropic
(25/50/25), and soft (10/80/10) lay-up sequences, as well as a wide range of loading modes and
environmental conditions (CTD, RTA, ETD, and ETW). Most of these data sets contain
specimens from multiple distinct material batches. Shape parameters obtained from staticstrength distributions were combined for several different analysis scenarios to investigate the
degree to which these parameters affect the MSSP of AS4C-5HS (table 21). The distribution of
52
shape parameters calculated for AS4C-5HS using the static data in FAA-LVM is shown in
figure 29.
Table 21. Summary of Weibull Shape Parameter Analysis of AS4C-5HS
Weibull Statistics
Analysis Case
Analysis Variable
All
Test environment

2.104
2.263
2.284
1.976
2.926
1.985
2.834
2.062
2.643
2.454
2.293
AS4C-5HS
CTD
RTA
ETW
40/20/40
25/50/25
10/80/10
Lamina
All bearing
OH/FH
NH
Lay-up
Loading mode

35.694
37.711
36.714
34.586
29.475
36.759
46.860
26.579
21.535
41.192
42.576
Modal
26.267
29.144
28.534
24.207
25.549
25.828
40.190
19.266
17.990
33.279
33.163
Number of
Data Sets
78
14
26
34
13
18
16
20
5
29
16
100
Shape Parameter
80
60
40
20
0
0
10
40/20/40
20
30
25/50/25
40
50
10/80/10
60
70
Lamina
80
Figure 29. Shape Parameters for AS4C-5HS Static-Strength Distributions
53
90
4.1.7 Nelcote T700/E765 Graphite Unidirectional Tape.
Nelcote (formally FiberCote) T700/E765 24K graphite unidirectional tape (E765-UT) material
from the FAA-LVM material database [33] are analyzed in this section. This database contains
834 E765-UT specimens from 47 data sets. These data sets represent hard (50/40/10), quasiisotropic (25/50/25), and soft (10/80/10) lay-up sequences, as well as a wide range of loading
modes and environmental conditions (CTD, RTA, and ETW). Most of these data sets contain
specimens from multiple distinct material batches. Shape parameters obtained from staticstrength distributions were combined for several different analysis scenarios to investigate the
degree to which these parameters affect the MSSP of E765-UT (table 22). The distribution of
shape parameters calculated for E765-UT using the static data in FAA-LVM is shown in
figure 30.
Table 22. Summary of Weibull Shape Parameter Analysis of E765-UT
Weibull Statistics
Analysis Case
Analysis Variable
All
E765-UT
Test environment CTD
RTA
ETW
Lay-up
40/20/40
25/50/25
10/80/10
Loading mode
All bearing
OH/FH
VNRS
Unnotched

2.0867
2.241
2.117
1.779
1.976
2.158
2.445
3.280
2.343
3.229
2.222

30.719
22.254
31.091
23.739
32.334
29.647
30.220
31.484
39.705
14.368
23.345
Number of
Modal Data Sets
22.471
47
17.095
7
22.98
29
14.920
7
22.627
16
22.215
20
24.372
8
28.180
16
31.312
14
12.810
4
17.838
9
80
Shape Parameter
60
40
20
0
0
10
50/40/10
20
30
25/50/25
40
50
10/80/10
60
Figure 30. Shape Parameters for E765-UT Static-Strength Distributions
54
4.1.8 Nelcote T300/E765 3K Plain-Weave Fabric.
Nelcote (formally FiberCote) T700/E765 plain-weave fabric (E765-PW) material from the FAALVM material database [33] are analyzed in this section. This database contains 722 E765-PW
specimens from 48 data sets. These data sets represent hard (40/20/40), quasi-isotropic
(25/50/25), and soft (10/80/10) lay-up sequences, as well as a wide range of loading modes and
environmental conditions (CTD, RTA, and ETW). Most of these data sets contain specimens
from multiple distinct material batches. Shape parameters obtained from static-strength
distributions were combined for several different analysis scenarios to investigate the degree to
which these parameters affect the MSSP of E765-PW (table 23). The distribution of shape
parameters calculated for E765-PW using the static data in FAA-LVM is shown in figure 31.
Table 23. Summary of Weibull Shape Parameter Analysis of E765-PW
Weibull Statistics
Analysis Case
Analysis Variable
All
Test environment
E765-PW
CTD
RTA
ETW
40/20/40
25/50/25
10/80/10
All bearing
OH/FH
VNRS
NH
Lay-up
Loading mode

2.389
2.434
2.535
2.751
2.484
2.947
1.907
2.200
3.098
1.288
4.619

32.735
24.535
35.837
27.714
37.800
30.194
27.820
30.491
32.641
26.488
38.055
Number of
Data Sets
Modal
26.089
19.740
29.400
23.516
30.723
26.233
18.843
23.148
28.782
8.269
36.097
48
7
31
7
17
20
8
16
15
4
9
80
Shape Parameter
60
40
20
0
0
10
40/20/10
20
30
25/50/25
40
50
10/80/10
60
Figure 31. Shape Parameters for E765-PW Static-Strength Distributions
55
4.1.9 Adhesive Effects of Defects Data.
The FAA-EOD database [36] contains more than 70 bonded-joint PFS element tests that have
disbonds, lightening strikes, and low-velocity impact damages. These specimens were fabricated
using T700-PW and 7781-8HS and bonded with the EA9394 two-part paste adhesive system.
Only data sets that contain more than five specimens were included in the Weibull analysis
(table 24).
Table 24. Weibull Parameters for Bonded-Joint PFS Element Tests
Adherend Material
T700-PW
7781-8HS
Defect Description
No disbonds
Small disbonds (circle, diamond)
Large rectangular disbonds
Lightning strikes
Low-velocity impact damages
No disbonds
Shape
Parameter, α̂
11.358
18.319
14.181
22.390
20.255
19.701
Number of
Specimens
9
6
9
6
20
5
In addition to the above data, 60 bonded joints (elements), impacted at different energy levels
[36] and tested in SLS, were analyzed. Specimens with a 4- by 4-inch gage section were
fabricated using the following material systems (commonly used in aircraft applications) and
bonded using EA9394 two-part paste adhesive system:



Newport NB321/7781 E-glass satin weave (FGSW)
Toray T700G-12K/3900-2 carbon fabric plain weave (CFPW)
Toray T800S/3900-2B carbon tape unidirectional (CTU)
Although three different impactor diameters (0.50, 0.75, and 1.00 inch) and three different
energy levels (88.5, 221, and 354 in-lbf) were used to inflict damage, the impact damages were
contained mostly within the elastic trough and away from the side edges. Thus, although the
damage states (i.e., residual indentation and damage area) were different, the residual strength
was not significantly influenced by the impact parameters. Thus, the data for each material were
pooled and analyzed for scatter (table 25).
Table 25. Weibull Parameters for Bonded SLS Element Tests
Weibull Statistics
Adherend
Adhesive
FGSW
CFPW
CTU
EA9394
α̂ 
14.134
25.657
19.890
56
β̂ 
(lbf)
847.347
867.356
986.267
Number of
Specimens
20
20
20
4.2 SUMMARY.
The databases used for the strength scatter analysis included coupon- and component-level staticstrength data for different lay-up configurations, test environments, and loading modes, and
included solid laminates, sandwich construction, and bonded joints. First, the shape parameters
obtained for different strength data sets were pooled separately by material system and test
environment, and compared to strength shape parameters obtained by pooling all the data
(denoted by All) for each material system. Figure 32 shows that ETW data have the highest
scatter while RTA data have the least scatter, for most cases analyzed in this section. The scatter
in ETW data can be attributed to variations in total moisture absorption among test specimens
and the time to reach the elevated test temperature prior to test. Furthermore, the shape
parameter obtained by pooling test data from all environmental conditions is close to the shape
parameter obtained by analyzing RTA test data of each material system. Also, the pooled values
are higher than the MSSP of 20, which was the value used for U.S. Navy F/A-18 certification,
for all the material systems analyzed in this report. This can be attributed to the improvements in
materials, process techniques, and test methodologies of modern composite materials.
40
CTD
35
RTA
Static Strength Shape Parameter
ETW
30
All
25
20
15
10
5
0
AS4/E7K8
PW
T700/#2510
UNI
T700/#2510
PW
7781/#2510 AS4C/MTM45 AS4C/MTM45 T700/E765
8HS
UNI
5HS
UNI
T300/E765
PW
Material System
Figure 32. Comparison of Composite Strength Shape Parameters for Different Environments
Second, the previous exercise was repeated by pooling data by different lay-ups. Figure 33
indicates that the shape parameters of T700-PW and E765 unidirectional lay-up (UNI) are
independent of lay-up sequence, while other material systems indicate large variations among
lay-up sequences. Also, the shape parameter obtained by pooling test data from all laminate
stacking sequences is close to the shape parameter obtained by analyzing the quasi-isotropic
laminate (25/50/25) test data of most of the material systems. For both ACG material systems
(AS4C-UT and AS4C-5HS), the soft laminate (10/80/10) test data indicate the least scatter while
T700-UNI and E765-PW indicate the most scatter. The hard laminate (50/40/10) test data
indicate the reverse trend for these four material systems.
57
45
Static Strength Shape Parameter
40
50/40/10
25/50/25
35
10/80/10
All
30
25
20
15
10
5
0
AS4/E7K8
PW
T700/#2510
UNI
T700/#2510
PW
7781/#2510 AS4C/MTM45 AS4C/MTM45
8HS
UNI
5HS
T700/E765
UNI
T300/E765
PW
Material System
Figure 33. Comparison of Composite Strength Shape Parameters for Different Lay-Ups
Third, the data were pooled by specimen configuration. Figure 34 shows that the loading mode
significantly influences the static scatter for all five material systems. For both Nelcote material
systems (E765-UNI and E765-PW), unnotched (NH) test data indicate the most scatter, resulting
in shape parameters significantly lower than 20, while OH and FH data indicate the least scatter.
For both ACG material systems, bearing test data indicate the most scatter. For both Toray
material systems (T700-UNI and T700-PW), OH and FH data indicate the most scatter, close to
the scatter in unnotched test data, while bearing test data indicate the least scatter.
45
Bearing
40
Static Strength Shape Parameter
OH/FH
35
NH
All
30
25
20
15
10
5
0
AS4/E7K8
PW
T700/#2510
UNI
T700/#2510
PW
7781/#2510
8HS
AS4C/MTM45 AS4C/MTM45
UNI
5HS
T700/E765
UNI
T300/E765
PW
Material System
Figure 34. Comparison of Composite Strength Shape Parameters for Different Loading Modes
58
The scatter analysis results in this section provide guidance to design a cost-effective test matrix
for generating reliable MSSP. The data are documented so they can be readily available for a
particular case, and the user does not have to generate new data or analyze the scatter. To
generate a life shape parameter that represents the general scatter of current typical aircraft
composite materials, the shape parameters representing all composite and adhesive strength data
sets included in this report are pooled. However, the authors recommend generating shape
parameters based on the materials, processes, and design details related to the critical areas of a
particular structure rather than using pooled data to ensure safety and reliability of fatigue life
predictions based on scatter analysis. This exercise was done to establish a generic scatter
distribution of modern composite strength data and to compare that against the values proposed
by Whitehead, et al. [2]. The resulting (modal) strength shape parameter or R is 26.31.
5. FATIGUE LIFE DATA SCATTER ANALYSIS.
An extensive fatigue scatter analysis was conducted by Whitehead, et al. [2], on various material
databases. These databases represented several structural details and variables such as R-ratio,
laminate lay-up, loading mode, load transfer, specimen geometry, and test environment. The
material database of Badaliance and Dill [43] included 204 graphite/epoxy data sets, whereas
Whitehead and Schwarz [45] included 2925 data points from 120 data sets of graphite/epoxy,
450 data points from 26 data sets of E-glass/epoxy, and 419 step-lap bonded joints from 23 data
sets. To increase the accuracy of the Weibull analysis, only the data sets containing five or more
specimens were included in the individual Weibull analysis.
Typically, an S/N curve contains fatigue failures for multiple stress levels in addition to staticstrength data points and any run-outs. In this report, the fatigue life scatter was analyzed using
individual Weibull, joint Weibull, and Sendeckyj analyses. Only data sets containing more than
five specimens within a stress level were included in the individual Weibull analysis. Residual
strength data for all run-outs were included in the Sendeckyj analysis.
5.1 STRUCTURAL DETAILS FOR FATIGUE LIFE SCATTER ANALYSIS.
This section includes over one thousand data points from the following composite material
systems commonly used in aircraft applications:



AS4/E7K8 plain-weave fabric (AS4-PW)
T700/#2510 plain-weave fabric (T700-PW)
7781/#2510 8-harness satin-weave fabric (7781-8HS)
In addition, adhesive fatigue data from the FAA-D5656 material database [35] in the following
test environments for three adhesive systems are included:



Hysol EA9696 film adhesive
PTM&W ES6292 paste adhesive
Cessna Aircraft proprietary paste adhesive (Loctite)
59
Data generated from coupons and elements were used to investigate the dependence of the
fatigue shape parameter, which is a representation of fatigue-life scatter, on various coupon
geometries, loading modes, environments, and lay-ups. The degree to which these parameters
affect the overall LEF factors from a parametric basis is discussed in section 5.4. Since fatiguelife data inherently exhibit significantly more scatter than static-strength data, the MLSP, or L,
is noticeably smaller than the MSSP, or R. Several examples are shown for obtaining MLSP by
pooling different data sets. When pooling data to estimate MLSP, the user is advised to select
appropriate design details that are applicable to a certain structure.
The material databases described in this section represent typical examples of commonly used
composite material systems. These materials are used in several ongoing certification programs
and certified general aviation aircraft. The databases include coupon-level static-strength data
for the following primary variables:

Lay-ups—hard, quasi-isotropic, and soft (typically 50/40/10, 25/50/25, and 10/80/10,
respectively, for unidirectional material, and 40/20/40, 25/50/25, and 10/80/10,
respectively, for fabric material) and all ±45° plies (0/100/0)

Environments—CTD, RTA, RTW, ETW

Tension—OH

Compression—OH

Interlaminar shear—DNC
In addition to coupon-level data, element-level static-strength data were included in the analysis
to represent the following design details/requirements:



Sandwich—core materials, facesheet
Bonded joints—SLS
Damage tolerance—CAI, sandwich
The following sections include the generation of shape parameters for several different material
systems accounting for the effects of different geometries, environments, lay-ups, and loading
modes.
5.2 FATIGUE SCATTER ANALYSIS.
ReliaSoft Weibull software and SACC were used for the Sendeckyj analysis of fatigue data in
two steps: with and without static-strength data. Residual strength data of all runout specimens
were included in the Sendeckyj analysis. Individual and joint Weibull analyses were conducted
using only the fatigue data. Kassapoglou [31] life predictions based only on static-strength
scatter were compared with Sendeckyj analysis of experimental data in appendix A. Sendeckyj
fitting parameters were recalculated for each S/N data set using SACC to generate the Sendeckyj
fitting curves. As shown in appendix A, currently Kassapoglou methodology either under- or
60
overpredicts the fatigue life significantly for some S/N curves. Note that Kassapoglou
methodology predicts the fatigue life based only on static data distribution.
5.2.1 The AS4/E7K8 Plain-Weave Fabric.
Fatigue analysis of 385 AS4-PW specimens from 14 data sets is included in this section. Each
data set contains a minimum of three fatigue stress levels and at least six static-strength data
points (figure 35). Figure 35 includes the static data, fatigue data (failures and run-outs) and the
residual strength data for run-outs. Once the Sendeckyj fitting parameters C and S are
determined, the fitting curve for the S/N data is obtained as shown in figure 35. In addition, the
equivalent static strength for each fatigue data point (failures and run-outs) is displayed. S/N
curves for AS4-PW are included in appendix A. Fatigue-life scatter analysis data are shown in
table 26. The shape parameters corresponding to each S/N curve using Sendeckyj, individual
Weibull, and joint Weibull scatter analysis are denoted as α̂ Sendeckyj , α̂ IW , and α̂ JW , respectively.
50000
Equivalent Static
Strength for Fatigue
Data Calculated using
Sendeckyj
Static
45000
40000
Stress (psi)
35000
Residual
Strength
30000
25000
Fatigue
Failures
Run-outs
20000
15000
Experiment
10000
Residual Strength
Equivalent Static Sterngth
5000
Sendeckyj
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure 35. Sendeckyj Wearout Analysis Prediction of Fatigue Life of OHT (R = 0) – AS4-PW
Figure 36 shows a comparison of AS4-PW OHC fatigue data, where R = -1 for quasi-isotropic
(25/50/25) and two resin-dominant lay-up configurations (10/80/10 and 0/100/0). The same data
are normalized with respect to the ultimate static strength (average) and are shown in figure 37.
Both figures show that the 0/100/0 lay-up with all ±45° plies exhibits the critical fatigue life but
not the data set that has the highest scatter.
61
Table 26. Fatigue-Life Scatter Analysis for AS4-PW
Specimen Configuration
Sendeckyj
α̂ Sendeckyj
Lay-Up
10/80/10
0/100/0
Sandwich
25/50/25
40/20/40
*
#
Test Description
OH
OHC
OHT
OH
CAI-BVID (20 ply)
CAI-VID (40 ply)
DNC
DNC
OH
TAI-BVID
TAI-VID
4PB-HRH 10
OH
CAI-BVID#
CAI-VID#
CAI-LID#
CAI-VID
R-Ratio
-1
5
0
-0.2
5
5
-1
-0.2
-1
0
0
0
-1
5
5
5
5
α̂ Sendeckyj
(with static) (without static)
2.068
2.604
1.792
2.328
3.434
3.686
3.319
4.090
2.870
3.321
2.103
2.221
3.837
3.905
2.025
1.962
2.495
3.480
1.640
1.515
1.111
1.065
1.924
2.020
3.224
3.661
1.774
2.234
2.182
2.658
2.466
2.799
2.337
3.522
Weibull*
α̂ IW 
α̂ JW
3.304
3.223
5.555
4.003
3.968
2.778
6.636
2.278
5.528
2.477
1.974
5.131
5.713
2.446
2.991
3.272
4.392
2.731
3.329
4.641
3.962
3.808
2.625
5.066
2.245
4.281
2.144
1.931
2.795
4.745
2.355
2.778
2.948
3.687
Life scatter analysis using Weibull only includes fatigue data.
Test data not included in the combined analysis for MLSP generation.
BVID = Barely visible impact damage
VID = Visible impact damage
LID = Large impact damage
Figure 38 shows a comparison of AS4-PW OH measured data for different R-ratios, and figure
39 shows a comparison of normalized data. Both figures indicate that R = -1 is the critical
fatigue life and has the least scatter when pooled with static data points.
62
-45
-40
-35
Max S tress (ksi)
-30
-25
-20
AS4/E7K8 PW [R = -1]
Ult. (OHC)
25/50/25 - -45.375 ksi
10/80/10 - -40.472 ksi
0/100/0 - -21.875 ksi
-15
-10
-5
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 36. Effects of Lay-Up Sequence, AS4/E7K8, OH Measured Fatigue Data
100
% of OH (ult)
80
60
40
AS4/E7K8 PW [R = -1]
Ult. (OHC)
25/50/25 - -45.375 ksi
10/80/10 - -40.472 ksi
0/100/0 - -21.875 ksi
20
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 37. Effects of Lay-Up Sequence, AS4/E7K8, OH Normalized Fatigue Data
63
45
35
AS4/E7K8 PW
Ult. (OH)
OHC - -40.472 ksi
OHT - 43.455 ksi
R=0
R=5 R=
R-0.2
= -0.2
Max. S tress (ksi)
25
15
5
R = -1
-5 1
10
100
1000
10000
100000
1000000
-15
-25
-35
-45
Cycles
Figure 38. Effects of Stress Ratio for AS4-PW OH Measured Fatigue Data
100
% of OH (ult)
80
60
40
AS4/E7K8 PW
Ult. (OH)
OHC - -40.472 ksi
OHT - 43.455 ksi
R=0
= -0.2
R = 5 R =R-0.2
20
R = -1
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 39. Effects of Stress Ratio for AS4-PW OH Normalized Fatigue Data
64
Figure 40 shows a comparison of AS4-PW CAI data for different lay-ups and impact energy
levels. Normalized data indicate that the fatigue life is independent of lay-up, up to
1500 in-lbf/in impact energy. Figure 41 shows a comparison between CAI and TAI for different
energy levels. Normalized data indicate that the fatigue life for TAI is independent of impact
energy level up to 1500 in-lbf/in and is significantly lower than for CAI. The lower fatigue life
for TAI can be attributed to the fact that these specimens are all ±45°, while the CAI specimens
are 10/80/10 lay-up. TAI data also indicate significant scatter compared to CAI data.
100
% of CAI (ult)
80
60
40
AS4/E7K8 PW
10/80/10 -- Ult. (CAI)
20 plies (500) -34.605 ksi
40 plies (1500) -30.263 ksi
40/20/40 -- Ult (CAI)
20
20 plies (1500) -31.845 ksi
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 40. Effects of Lay-Up Sequence for AS4-PW CAI Normalized Fatigue Data
100
% of CAI/TAI (ult)
80
60
40
AS4/E7K8 PW
10/80/10 -- Ult. (CAI)
20 plies (500) -34.605 ksi
0/100/0 -- Ult (TAI)
20 plies (500) 17.799 ksi
20
20 plies (1500) 15.118 ksi
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 41. Comparison of CAI and TAI for AS4-PW
65
Figure 42 shows the summary of a detailed fatigue life scatter analysis conducted on AS4-PW
using individual Weibull, joint Weibull, and Sendeckyj analyses (with and without static data in
the S/N curves). These three techniques are also used in reference 2 for generating life and
LEFs. However, the fatigue life scatter analysis for deriving life and LEFs is not limited to these
three techniques. The Sendeckyj analysis includes runout and corresponding residual strength
data points. Also, the Sendeckyj analysis produces a higher scatter as it is unbiased compared to
the arithmetically averaged individual Weibull shape parameters (figure 42). Overall, the
individual Weibull analysis provides the highest fatigue-life shape parameter, while the
Sendeckyj analysis provides the lowest (conservative) fatigue-life shape parameter. For most
cases, when the static data are included in the Sendeckyj analysis, the analysis resulted in
significantly lower fatigue-life shape parameters than that without the static test data. For
example, 7781-8HS OHT (R = 0) S/N data resulted in a fatigue-life shape parameter of 9.794 for
Sendeckyj analysis without static data. When the static test data are included in the analysis, this
value was reduced to 1.769. For most composite analysis cases, the Sendeckyj analysis without
static test data and the joint Weibull analysis resulted in similar shape parameters.
7.0
Sendeckyj (w / static)
Sendeckyj (w /o static)
Individual Weibull
6.0
5.0
4.0
3.0
2.0
10/80/10
0/100/0
HRH 10
Sandw ich
25/50/25
CAI - VID (R=5)
CAI - LID
(R=5)#
CAI - VID
(R=5)#
CAI - BVID
(R=5)#
OH (R=-1)
Flex (R=0)
TAI - VID (R=0)
TAI - BVID
(R=0)
OH (R=-1)
DNC (R=-0.2)
DNC (R=-1)
CAI - VID (R=5)
CAI- BVID
(R=5)
OH (R=-0.2)
OHT (R=0)
0.0
OHC (R=5)
1.0
OH (R=-1)
Fatigue Life Weibull Shape Parameter
Joint Weibull
40/20/40
[OH (T/C): open-hole (tension/compression), CAI: compression after impact, TAI: Tension after impact, Flex: fourpoint bend flexure, DNC: double-notched compression, BVID: barely visible impact damage, VID: visible impact
damage, LID: large impact damage]
Figure 42. Fatigue-Life Shape Parameters of AS4-PW From Different Analysis Methods
The Sendeckyj analysis is the most conservative of the above-mentioned methods because it
includes static, fatigue failures, run-outs, and residual strength data. This analysis produces the
equivalent static strength for each fatigue data point. These Sendeckyj equivalent strength values
based on the fatigue data are pooled and compared against the tested static-strength data in table
27 and are shown in figure 43. Table 27 shows the coefficient of variation (CV) of each data set,
i.e., static or pooled Sendeckyj equivalent static-strength data.
66
Table 27. Comparison of Static Strength of AS4-PW From Test and Sendeckyj Analysis
Specimen Configuration
Lay-Up
10/80/10
0/100/0
Sandwich
25/50/25
40/20/40
Test Description
OH
OHC
OHT
OH
CAI-BVID (20 ply)
CAI-VID (40 ply)
DNC
DNC
OH
TAI-BVID
TAI-VID
4PB-HRH 10
OH
CAI-BVID
CAI-VID
CAI-LID
CAI-VID
R-Ratio
-1
5
0
-0.2
5
5
-1
-0.2
-1
0
0
0
-1
5
5
5
5
Static Test
Strength
CV
(ksi)
40.472
3.882
43.455
1.778
34.605
30.263
3.988
3.988
21.875
17.799
15.118
0.145
45.375
36.025
29.671
25.445
31.845
4.454
2.690
4.022
4.022
1.600
2.172
4.143
2.071
3.579
2.361
3.003
2.721
3.223
Pooled (Sendeckyj)
Strength
CV
(ksi)
42.608
3.848
41.360
2.783
42.823
4.217
42.942
5.055
35.581
3.238
30.329
2.060
3.975
4.456
4.008
5.843
22.058
3.420
16.709
9.695
14.807
5.112
0.141
8.411
45.882
2.499
36.289
2.002
29.990
2.438
25.614
2.259
31.586
3.945
BVID = Barely visible impact damage
VID = Visible impact damage
LID = Large impact damage
The fatigue-life shape parameters for AS4-PW shown in table 26 are then fitted into another
Weibull distribution, and the shape parameters corresponding to the new distribution and the
model shape parameter are obtained from three different techniques: MLE, RRX, and RRY. As
shown in table 28, there were no significant differences between the results of the different
techniques. The Sendeckyj-model shape parameter (with static data) for life scatter was 2.427,
while it was 3.974 for individual Weibull.
67
50
12.0
Stength (Static)
45
Stength (Pooled)
10.0
CV (Static)
40
CV (Pooled)
35
6.0
20
4.0
15
10
2.0
10/80/10
0/100/0
HRH-10
25/50/25
CAI - VID (R=5)
CAI - LID (R=5)
CAI - VID (R=5)
0.0
CAI - BVID (R=5)
OH (R=-1)
Flex (R=0)
TAI - VID (R=0)
TAI - BVID (R=0)
OH (R=-1)
DNC (R=-0.2)
DNC (R=-1)
CAI - VID (R=5)
CAI- BVID (R=5)
OH (R=-0.2)
OHT (R=0)
0
OHC (R=5)
5
40/20/40
BVID = Barely visible impact damage
VID = Visible impact damage
LID = Large impact damage
Figure 43. Comparison of Static Strength and CV of AS4-PW From Test and
Sendeckyj Analysis
Table 28. Weibull Statistics for Combined Distribution of AS4-PW
Analysis
Method
MLE
RRX
RRY
Weibull
Parameter

Fatigue Scatter Analysis Method
Sendeckyj
Individual
Joint
Sendeckyj
(with static) (without static) Weibull Weibull
3.553
3.511
3.207
3.889

2.716
3.138
4.558
3.800
Modal
2.475
2.852
4.056
3.520

3.443
2.901
3.065
3.714

2.710
3.163
4.541
3.786
Modal
2.453
2.734
3.992
3.479

3.323
2.798
2.953
3.547

2.725
3.184
4.572
3.811
Modal
2.447
2.719
3.974
3.472
68
Coeff. Variation (%)
25
OH (R=-1)
Strength (ksi)
8.0
30
5.2.2 The T700/#2510 Plain-Weave Fabric.
Fatigue analysis of 240 T700-PW specimens from seven S/N curves is included in this section.
Each data set contains a minimum of three fatigue stress levels and at least six static-strength
data points. Similar to AS4-PW, T700-PW OH data show that R = -1 is the most critical stress
ratio for both measured (figure 44) and normalized (figure 45) data comparisons. Fatigue-life
scatter analysis data are shown in table 29. Figure 46 shows a comparison of fatigue-life scatter
data obtained from different analysis techniques. Similar to AS4-PW, Sendeckyj analysis
produces a higher scatter for T700-PW, as it is unbiased compared to the arithmetically averaged
individual Weibull shape parameters. However, Sendeckyj data exhibit significantly less scatter
compared to AS4-PW. As a result, the difference between Sendeckyj and individual Weibull
fatigue-life shape parameters is not pronounced for T700-PW (table 30).
45
35
Max. S tress (ksi)
25
15
5
-5 1
T700/#2510 PW
Ult. (OH)
OHC - -34.986 ksi
OHT - 41.686 ksi
R=0
R= 5 RR
= -0.2
= -0.2
R = -1
10
100
1000
10000
100000
1000000
-15
-25
-35
-45
Cycles
Figure 44. Effects of Stress Ratio for T700-PW OH Measured Fatigue Data
69
100
% of OH (ult)
80
60
40
T700/#2510 PW
Ult. (OH)
OHC - -34.986 ksi
OHT - 41.686 ksi
R=0
R = 5 R =R-0.2
= -0.2
R = -1
20
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 45. Effects of Stress Ratio for T700-PW OH Normalized Fatigue Data
Table 29. Fatigue-Life Scatter Analysis for T700-PW
Specimen Configuration
Lay-Up
10/80/10
Test
Description
OH
OHC
OHT
α̂ Sendeckyj
R-Ratio
-1
α̂ Sendeckyj
(with static) (without static)
2.796
3.389
Weibull
α̂ IW 
α̂ JW
3.932
3.793
5
0
1.481
1.367
1.992
2.086
2.344
2.721
2.285
2.207
-0.2
1.904
2.079
2.615
2.320
DNC
-1
1.496
1.464
1.714
1.765
DNC
-0.2
1.547
1.635
2.489
1.949
0
1.764
2.315
2.881
2.464
OH
Sandwich
Sendeckyj
4PB-HRH 10
70
4.5
4.0
Sendeckyj (w/ static)
Sendeckyj (w/o static)
Individual Weibull
Fatigue Life Weibull Shape Parameter
3.5
Joint Weibull
3.0
2.5
2.0
1.5
1.0
0.5
0.0
OH (R = -1)
OHC (R = 5)
OHT (R = 0)
OH (R = -0.2)
DNC (R = -1)
DNC (R = -0.2)
10/80/10
Flex (R = 0)
HRH 10
Sandwich
Figure 46. Fatigue-Life Shape Parameters of T700-PW From Different Analysis Methods
Table 30. Weibull Statistics for Combined Distribution of Life Distributions of T700-PW
Analysis
Method
Weibull
Parameter
MLE

RRX
RRY
Fatigue Scatter Analysis Method
Sendeckyj
Individual
Sendeckyj
(with static) (without static) Weibull
3.795
3.728
4.433
Joint
Weibull
3.825

1.944
2.359
2.919
2.638
Modal
1.793
2.170
2.756
2.437

5.255
4.344
4.570
5.085

1.892
2.326
2.908
2.578
Modal
1.817
2.190
2.755
2.470

3.780
3.729
4.145
3.905

1.965
2.371
2.941
2.658
Modal
1.811
2.181
2.752
2.464
5.2.3 The 7781/#2510 8-Harness Satin-Weave Fabric.
Fatigue analysis of 204, 7781-8HS specimens from seven S/N curves is included in this section.
Each data set contains a minimum of three fatigue stress levels and at least six static-strength
data points. As shown for the previous two material systems, OH with R = -1 data indicate the
lowest fatigue life (figure 47). Normalized OH data for R = 5 indicate the lowest level of fatigue
degradation for this material (figure 48). Fatigue-life scatter analysis data are shown in table 31.
71
35
25
7781/#2510 FG
Ult. (OH)
OHC - -33.227 ksi
OHT - 26.808 ksi
R=0
R = 5 R =R-0.2
= -0.2
Max. S tress (ksi)
15
5
R = -1
-5 1
10
100
1000
10000
100000
1000000
-15
-25
-35
Cycles
Figure 47. Effects of Stress Ratio for 7781-8HS OH Measured Fatigue Data
100
% of OH (ult)
80
60
40
7781/#2510 FG
Ult. (OH)
OHC - -33.227 ksi
OHT - 26.808 ksi
R=0
R -0.2
= -0.2
R=5 R=
20
R = -1
0
1
10
100
1000
Cycles
10000
100000
1000000
Figure 48. Effects of Stress Ratio for 7781-8HS OH Normalized Fatigue Data
72
Table 31. Fatigue-Life Scatter Analysis for 7781-8HS
Specimen Configuration
Test
Description
OH
Lay-Up
10/80/10
R-Ratio
-1
Weibull
α̂ Sendeckyj
α̂ Sendeckyj
OHC
OHT
(with static) (without static)
1.876
1.730
α̂ IW 
α̂ JW
6.258
5.139
5
0
2.727
1.769
2.580
9.794
3.169
13.866
2.668
12.667
-0.2
1.575
7.335
8.998
7.387
DNC
-1
1.163
1.056
2.035
1.711
DNC
-0.2
2.724
2.643
3.346
2.760
0
2.358
3.429
4.034
3.441
OH
Sandwich
Sendeckyj
4PB-HRH 10
Figure 49 shows a comparison of fatigue-life scatter data obtained from different analysis
techniques. Similar to AS4-PW and T700-PW, Sendeckyj analysis produces a higher scatter for
7781-8HS than individual Weibull shape parameters. Except for R = 5, OH data without static
data show significantly high shape parameters, indicating an unrealistic skewness in OH fatigue
data. Table 32 includes a summary of the fatigue-life parameter for 7781-8HS.
Sendeckyj (w/ static)
16.0
Sendeckyj (w/o static)
Individual Weibull
14.0
Joint Weibull
Weibull Shape Parameter
12.0
10.0
8.0
6.0
4.0
2.0
0.0
OH (R = -1)
OHC (R = 5)
OHT (R = 0)
OH (R = -0.2)
10/80/10
DNC (R = -1)
DNC (R = -0.2)
Flex (R = 0)
HRH 10
Sandwich
Figure 49. Fatigue-Life Shape Parameters of 7781-8HS From Different Analysis Methods
73
Table 32. Weibull Statistics for Combined Distribution of 7781-8HS
Analysis
Method
Weibull
Parameter

MLE
RRX
RRY
Fatigue Scatter Analysis Method
Sendeckyj
Sendeckyj
Individual
(with static) (without static)
Weibull
4.232
1.459
1.662
Joint
Weibull
1.574

2.237
4.542
6.726
5.748
Modal
2.099
2.056
3.864
3.028

3.519
1.441
1.698
1.663

2.251
4.482
6.614
5.622
Modal
2.047
1.971
3.918
3.234

3.410
1.338
1.554
1.502

2.262
4.605
6.801
5.809
Modal
2.043
1.647
3.501
2.801
5.2.4 Adhesive Fatigue Data.
Fatigue scatter analysis of the FAA-D5656 material database [35], which contains
ASTM D 5656 single-lap shear adhesive joints from four different adhesive systems, is included
in this section. This database contains 390 adhesive specimens from 12 S/N curves, which
represent three different test environments: CTD, RTA, and RTW. Each data set contains a
minimum of three fatigue stress levels and at least nine data points in each stress level in addition
to static-strength data. Fatigue-life scatter analysis data are shown in table 33. Figure 50 shows
a comparison of fatigue-life scatter data obtained from different analysis techniques.
Table 33. Fatigue-Life Scatter Analysis for FAA-D5656
Specimen Configuration
Adhesive
Loctite
EA9696
PTM&W (0.06″)
PTM&W (0.16″)
Test
Environment
CTD
RTA
RTW
CTD
RTA
RTW
CTD
RTA
RTW
CTD
RTA
RTW
Sendeckyj
α̂ Sendeckyj
α̂ Sendeckyj
(with static)
0.805
0.662
0.682
0.847
0.403
0.379
0.870
1.051
0.681
0.363
0.856
0.671
(without static)
0.821
1.624
0.644
4.119
1.389
2.189
1.376
1.483
1.169
0.669
4.296
1.618
74
Weibull
α̂ IW 
α̂ JW
1.069
1.226
1.109
2.372
2.077
1.110
1.541
1.179
1.417
1.165
2.170
1.061
1.007
1.077
0.924
2.294
1.514
1.047
1.254
1.153
1.005
0.908
1.627
0.958
Individual Weibull shape and scale parameters of PTM&W for six shape parameters shown in
table 33 are 3.870 and 1.568, which result in a fatigue-life shape parameter of 1.451. Overall,
the Sendeckyj analysis exhibits high scatter, especially for the case with static strength,
compared to individual Weibull analysis. This is due to the scatter observed in adhesive fatigue
data at each stress level.
5.0
4.5
4.0
Sendeckyj (w/ static)
Sendeckyj (w/o static)
Individual Weibull
Fatigue Life Shape Parameter
3.5
Joint Weibull
3.0
2.5
2.0
1.5
1.0
0.5
0.0
CTD
RTD
Loctite
RTW
CTD
RTD
RTW
EA9696
CTD
RTD
PTM &W (0.06")
RTW
CTD
RTD
RTW
PTM &W (0.16")
Figure 50. Fatigue-Life Shape Parameters of FAA-D5656 Data From Different
Analysis Methods
Typically, the scatter in adhesive test data is significantly higher than that of composite test data.
This is primarily due to the processing parameters and the multiple secondary loading modes that
occur during testing of adhesive joints, i.e., peel stress at the ends of the overlap during singlelap shear tests. Postfatigue microscopic analysis of Loctite paste adhesive test specimens
indicated that fractures initiated from the clusters of glass beads, which were mixed into the
adhesive to control bondline thickness. Furthermore, the moisture-conditioned specimens
indicated localized swelling around the glass beads. The random distribution of glass beads and
the effects of the two above-mentioned phenomena resulted in large data scatter. The change in
bondline thickness and the amount of adhesive that squeezes out at the ends of the overlap causes
significant changes to the stress distribution of the adhesive layer, and consequently a large data
scatter. For the adhesive fatigue-life data in the current analysis, scatter significantly increases
with fatigue test times to failure, i.e., data from low stress levels indicated significant data
scatter. Therefore, individual Weibull analysis is recommended for adhesive test data. Also, the
modal shape parameters obtained for individual Weibull analysis data are consistent throughout
RRX, RRY, and MLE (table 34).
75
Table 34. Weibull Statistics for Combined Distribution of FAA-D5656 Data
Analysis
Method
Weibull
Parameter
MLE

RRX
RRY
Fatigue Scatter Analysis Method
Sendeckyj
Individual
Sendeckyj
(with static) (without static) Weibull
3.854
1.679
3.339
Joint
Weibull
3.141

0.764
2.016
1.626
1.371
Modal
0.707
1.176
1.461
1.213

3.374
1.993
4.487
4.833

0.765
1.944
1.572
1.320
Modal
0.689
1.371
1.486
1.258

3.064
1.741
3.247
3.426

0.777
2.020
1.644
1.381
Modal
0.683
1.237
1.468
1.249
As shown in figure 50, the Sendeckyj model with static data indicate significant scatter in the
adhesive data, whereas without static data, in some cases, scatter is higher than for individual
Weibull analysis. When including static data for the Sendeckyj analysis, it is recommended that
a minimum of six static data points be included. At this point, it is recommended that adhesive
S/N curves be analyzed using individual Weibull analysis. Thus, these S/N curves must have a
minimum of six data points in each stress level, and each S/N curve must have a minimum of
three stress levels in addition to static data.
5.3 SUMMARY OF FATIGUE SCATTER ANALYSIS.
Overall, the individual Weibull analysis provided the highest fatigue-life shape parameter, while
the Sendeckyj analysis provided the lowest (most conservative) fatigue-life shape parameter.
For most cases, when the static data are included in the Sendeckyj analysis, the analysis resulted
in significantly lower fatigue-life shape parameters than that without the static test data. For
example, 7781-8HS OHT (R = 0) S/N data resulted in a fatigue-life shape parameter of 9.794 for
Sendeckyj analysis without static data. When the static test data are included in the analysis, this
value was reduced to 1.769. For most composite analysis cases, the Sendeckyj analysis without
static test data and the joint Weibull analysis resulted in similar shape parameters.
Figure 51 shows a comparison of fatigue-life shape parameters calculated using the Sendeckyj
model for three composite material systems and three adhesive systems discussed in this section.
Sendeckyj analysis for each composite and adhesive S/N curve is conducted with and without
static data. Then, shape parameters for adhesive S/N curves are combined with shape parameters
for each composite system. Fatigue-life scatter with static and adhesive data indicate the highest
scatter for all three composite materials. Removing the adhesive data had the greatest effect on
7781-8HS life scatter, while removing the adhesive and static data had the greatest effect on
AS4-PW life scatter. For all three composites, removal of static and adhesive data resulted in
76
significantly less scattered life shape parameters. For both AS4-PW and T700-PW, the shape
parameter distribution obtained without static and adhesive data indicated the least scatter.
For composite material systems included in this section, R = -1 stress ratio resulted in the most
critical OH fatigue life. For T700-PW, this stress ratio resulted in the least scatter. Typically,
load reversal (R <0) causes low fatigue life and less scatter in test data.
Fatigue Life Shape Parameter
3.0
2.5
2.0
1.5
1.0
0.5
FAA-LEF
& FAA-Adhesive
AS4/E7K8
PW
FAA-LEF
& FAA-Adhesive
T700/#2510
PW
w/o static &
adhesive
w/o static &
w/ adhesive
w/ static &
w/o adheive
w/ static &
adhesive
w/o static &
adhesive
w/o static &
w/ adhesive
w/ static &
w/o adheive
w/ static &
adhesive
w/o static &
adhesive
w/o static &
w/ adhesive
w/ static &
w/o adheive
w/ static &
adhesive
0.0
FAA-LEF
& FAA-Adhesive
7781/#2510
8HS
Figure 51. Comparison of Fatigue-Life Shape Parameter for FAA-LEF Database
To generate a life shape parameter that represents the general scatter of current typically used
aircraft composite materials, the shape parameters representing all composite fatigue data sets
included in this report were pooled. However, the authors recommend generating shape
parameters based on the materials, processes, and design details related to the critical areas of a
particular structure rather than using pooled data to ensure safety and reliability of fatigue life
predictions based on scatter analysis. This exercise was done to establish generic scatter
distribution of modern composite fatigue data and to compare that against the values proposed by
Whitehead, et al. [2]. A summary of fatigue shape parameters obtained using different Weibull
analysis methods and different fatigue analysis techniques is shown in table 35. The resulting
MLSP (αL) is 2.13 (designated in the table with a dotted line), which was obtained using
Sendeckyj analysis (without static data in the S/N curves) for composite test data. It was
determined as 2.98 and 2.61 for individual and joint Weibull analyses, respectively. For an αL of
2.13, the life factor is 4.3. When the adhesive fatigue data are pooled with composite data, αL
was 1.60 using joint Weibull analysis, while it was 1.73 and 1.94 for Sendeckyj and individual
Weibull analyses, respectively. For an αL of 1.60, the life factor is 7.3.
77
Table 35. Weibull Statistics for Pooled Composite and Adhesive Data From Different
Analytical Techniques
Analysis
Data Set
Pooled composite
fatigue data
Analysis
Method
Weibull
Parameter
MLE
RRX
RRY
Pooled composite
and adhesive
fatigue data
MLE
RRX
RRY
Sendeckyj
(with static)
Sendeckyj
(without static)
Individual
Weibull
Joint
Weibull

3.342
1.826
1.826
1.853

2.408
3.291
4.602
3.974
Modal
2.165
2.131
2.980
2.614

3.900
2.674
2.801
3.105

2.371
3.150
4.362
3.752
Modal
2.198
2.644
3.725
3.310

3.656
2.320
2.285
2.432

2.394
3.251
4.561
3.942
Modal
2.193
2.550
3.546
3.170

2.070
1.698
1.565
1.586

1.968
2.917
3.726
3.229
Modal
1.430
1.729
1.943
1.723

2.000
2.206
2.130
2.167

1.961
2.826
3.556
3.095
Modal
1.387
2.150
2.641
2.326

1.951
2.063
1.888
1.938

1.975
2.876
3.679
3.190
Modal
1.366
2.086
2.466
2.194
5.4 LOAD-ENHANCEMENT FACTOR.
In this section, the LEF is calculated using the static strength and fatigue life distributions
derived from the data obtained in this study and historical data available at National Institute for
Aviation Research (NIAR). Data for three different composite material systems in the FAA-LEF
data set were combined with adhesive fatigue data in the FAA-D5656 material database to obtain
life factors and LEFs. As discussed in section 3.2, the life factor, NF, is a function of MLSP, or
L, and is not influenced by the MSSP, or R. However, LEF is a function of both parameters.
A combined load-life approach is also discussed.
Different combinations of MSSP and MLSP from sections 4.2 and 5.3, respectively, are
combined to calculate the corresponding LEF curves. Table 36 shows the life factors for MLSP
of composite data only and MLSP obtained by combining adhesive fatigue-life shape parameters
from individual Weibull analysis for adhesives and Sendeckyj analysis for composites (denoted
C and C+A, respectively, in table 36). MSSP for T700-PW also includes element test data from
the FAA-EOD database. The influence of test duration on the B-basis LEF for these three
materials is shown in figure 52 for one test article. This figure compares the LEF curves
generated for the data in this report against the LEF curves generated by Whitehead, et al. [2],
and Lameris [3] (referred to as NAVY and CASA, respectively). In addition, the strength and
78
MLSP obtained by pooling the data in this report (26.31 and 2.13, respectively) are used to
generate the LEF curve (referred to as NIAR for comparison in figure 52). Due to the significant
improvement in the MSSP used for the NIAR curve, the LEF requirements for lower test
durations are significantly reduced compared to both NAVY and CASA data. However, for test
durations of more than two lifetimes, CASA data resulted in lower LEF requirements than NIAR
data due to the lower MLSP obtained for CASA data. Note that CASA data include only 48
fatigue data points and the method of obtaining the fatigue shape parameter (i.e., Weibull or
Sendeckyj) is not specified. In contrast, NIAR data include over 800 composite fatigue data
points. Furthermore, the NIAR fatigue shape parameter selected for this curve is the most
conservative value (MLE and Sendeckyj in table 35) of the three fatigue analysis techniques
discussed in this report, corresponding to the largest life scatter. Nevertheless, NIAR data show
a significant reduction in LEF requirements compared to NAVY data, as shown in figure 52.
Table 36. Weibull Statistics for Combined Composites and Adhesives
Composite Material
AS4/E7K8 PW (C)
AS4/E7K8 PW (C+A)
T700/#2510 PW (C)
T700/#2510 PW (C+A)
7781/#2510 8HS (C)
7781/#2510 8HS (C+A)
α L
2.475
1.880
1.793
1.576
2.099
1.660
NF
3.431
5.267
5.752
7.516
4.364
6.715
Table 37 and figure 53 show a comparison of MLSP obtained for AS4-PW by using different
analytical techniques, i.e., individual Weibull or Sendeckyj model, for composites and adhesives.
AS4-PW data were analyzed with and without adhesive data (denoted C+A and C, respectively).
Composite data were analyzed using Sendeckyj method, while adhesive data were analyzed
using the individual Weibull method.
As observed in these examples, the LEF calculated as a function of test duration for these
materials is significantly lower than for NAVY. In addition, application of the life factor, rather
than LEF, for high loads in spectrum required fewer repetitions for improved MLSP than for
NAVY. However, guidance for generating reliable shape parameters must be established to
prevent unrealistic LEFs that compromise the structural integrity of composite aircraft. It is
recommended that specimens or elements representative of features of a particular structure (i.e.,
materials, design details, failure modes, loading conditions, and environments) be included in the
analysis rather than pooling various material databases. Also, it is noted that the primary goal in
scatter analysis is not selecting shape parameters from the critical lay-up, R-ratio, environment,
etc., (which may result in skewed data that will produce unconservative LEF), but rather
selecting the design details representing the critical areas of the structure.
79
1.200
NAVY
1.175
NIAR
CASA
1.150
AS4/E7K8 - C
AS4/E7K8 - C+A
T700/#2510 - C
1.125
T700/#2510 - C+A
7781/#2510 - C
LEF
7781/#2510 - C+A
1.100
1.075
1.050
1.025
1.000
1.0
2.0
4.0
8.0
Test Duration, N
NAVY
NIAR
Strength Shape Parameter
20.000
26.310
Life Shape Parameter
1.250
2.131
Life Factor
13.558
4.259
# of Lives (N)
LEF
1.00
1.177
1.125
1.50
1.148
1.088
2.00
1.127
1.063
2.50
1.111
1.044
3.00
1.099
1.029
4.25
1.075
1.000
5.00
1.064
0.987
8.00
1.034
0.950
13.60
1.000
0.908
Figure 52. Influence of Test Duration on B-Basis LEFs for Different Materials
80
Table 37. Weibull Statistics for AS4-PW Composites and Adhesives From Different
Analytical Techniques
Analysis Method
Individual Weibull (C)
Individual Weibull (C+A)
Sendeckyj (C)
Sendeckyj (C+A)
Sendeckyj (C)+Individual Weibull (A)
α L
4.056
2.082
2.475
1.021
1.880
NF
2.070
4.418
3.431
26.296
5.267
1.200
NAVY
1.175
NIAR
CASA
AS4 ~ Individual (C)
1.150
AS4 ~ Individual (C+A)
AS4 ~ Sendeckyj (C)
1.125
AS4 ~ Sendeckyj (C+A)
LEF
AS4 ~ Sendeckyj (C) +Individual (A)
1.100
1.075
1.050
1.025
1.000
1.0
10.0
Test Duration, N
Figure 53. Influence of Test Duration on B-Basis LEFs of AS4-PW From Different
Analytical Techniques
The primary objective of the analyses in sections 4 and 5 was to evaluate the parameters
affecting MSSP (R) and MLSP (L) so that minimum requirements to generate safe and
reliable, yet economical, LEFs and life factors (NF) can be outlined along with a
recommendation for benchmark test matrices. The secondary objective was to create a readily
available shared database of static-strength and fatigue-life shape parameters for commonly used
composite materials and structural details to support on-going and future certification programs,
thereby reducing testing time and cost. Finally, well-documented procedures and a user-friendly
computer code for generating statistically reliable life factor and LEFs were developed. Using
equation 9, tables 5 and 6 include A- and B-basis LEFs, respectively, for different combinations
of MSSPs and MLSPs for different test durations and numbers of test articles.
81
5.5 CASE STUDY—LIBERTY AEROSPACE XL2 FUSELAGE.
This section demonstrates an application of LEF on a composite fuselage of the Liberty
Aerospace XL2 aircraft (figure 54) that was primarily fabricated using T700/#2510 PW
(T700-PW) and Airex R82-80 foam core. To generate LEF for the fuselage material systems,
several sandwich coupon configurations were tested.
Figure 54. Liberty Aerospace XL2 Aircraft
A reduced test matrix was developed, including compression- and shear-loaded sandwich
coupons using an R82-80 core. Both notched and unnotched compression sandwich specimens
were tested using windowed anti-buckling fixtures. One-, two-, and three-ply facesheet lay-ups
and 3- and 5-mm-thick cores with and without core splices were included in the test matrix.
Several ETD static-compression specimens were also included. The total Liberty database
included 198 static data points and 82 fatigue data points. Each static data set had a minimum of
six specimens (table 38).
Table 38. Weibull Analysis Results of Liberty Sandwich Static Test Data
Specimen
Configuration
Loading
Configuration
RTA
unnotched
Compression
Compression
Compression
Compression
Compression
Compression
RTA
notched
Core Size
(mm)
Plies per
Facesheet
3
3
3
3
3
3
1
2
3
1
2
3
82
Core
Splice
Number of
Specimens
6
6
6
6
6
6
Weibull
Parameters
α̂
β̂
17.234
51.431
35.754
22.962
19.597
105.229
2.606
5.307
7.011
2.133
4.224
5.438
Table 38. Weibull Analysis Results of Liberty Sandwich Static Test Data (Continued)
Specimen
Configuration
Loading
Configuration
RTA shear
Shear
Shear
Shear
Shear
Shear
Shear
Compression
Compression
Compression
Compression
Compression
Compression
Compression
Compression
Compression
Compression
Compression
Compression
ETD static
Core Size
(mm)
Plies per
Facesheet
3
3
3
5
5
5
3
3
3
3
3
3
3
3
3
3
3
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
1
2
3
Core
Splice
Weibull
Parameters
Number of
Specimens
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
6
Yes
Yes
Yes
Yes
Yes
Yes
α̂
β̂
23.376
30.213
30.058
14.520
16.305
53.031
7.450
12.065
28.349
14.895
38.642
17.858
8.145
29.879
33.781
9.153
17.646
31.001
0.226
0.245
0.251
0.196
0.209
0.190
2.525
5.046
6.296
2.543
4.690
6.327
1.975
3.736
4.689
2.802
3.730
5.414
However, fatigue data sets did not have the minimum recommended five specimens per stress
level to generate accurate individual Weibull shape parameters. Therefore, MLSP was generated
using the Sendeckyj method for fatigue data sets that had a minimum of three stress levels, in
which at least two fatigue failures existed (table 39). MSSP was generated using several
scenarios: (1) only Liberty database [46]; (2) Liberty (without adhesive data) and FAA-LVM
databases; (3) Liberty (with adhesive data) and FAA-LVM databases; and (4) Liberty,
FAA-LVM, and FAA-EOD databases.
Table 39. Sendeckyj Analysis Results of Liberty Sandwich Fatigue Test Data
Specimen
Configuration
Unnotched
Notched
Loading
Configuration
Compression
Compression
Compression
Compression
Core Size
(mm)
3
5
3
5
Plies per
Facesheet
3
3
3
3
83
Core
Splice
Yes
Number of
Specimens
10
9
7
7
Shape
Parameter,
α̂
1.346
17.410
4.559
3.972
Table 39. Sendeckyj Analysis Results of Liberty Sandwich Fatigue Test Data (Continued)
Specimen
Configuration
Shear
Loading
Configuration
Shear
Shear
Shear
Shear
Shear
Shear
Core Size
(mm)
3
3
3
5
5
5
Plies per
Facesheet
1
2
3
1
2
3
Core
Splice
Shape
Parameter,
α̂
7.254
4.965
1.229
4.016
7.024
0.804
Number of
Specimens
9
8
9
7
7
9
Only 10/80/10 T700-PW (0/100/0 data was not available) and 50/40/10 T700-UT data from the
FAA-LVM database were included in the fatigue analysis of the Liberty XL2 fuselage materials
[46] because the Liberty XL2 fuselage was primarily fabricated using T700-PW with ±45° layup and highly orthotropic lay-ups of T700-UT around some of the highly loaded areas. This is
an example of using a shared database for generating reliable LEFs with minimum test efforts,
i.e., only testing the R82-80 core. B-basis LEFs are shown for all four analysis scenarios with
respect to different test durations in table 40. These data are also illustrated in figure 55.
Table 40. Comparison of Weibull Statistics for Liberty XL2 Database
Test Duration (N)
Analysis Scenario for Liberty XL2
1
2
3
4
1.166
1.141
1.148
1.157
1.158
1.131
1.138
1.146
1.147
1.118
1.124
1.132
1.133
1.100
1.106
1.112
1.121
1.085
1.090
1.095
1.110
1.073
1.076
1.081
1.079
1.035
1.037
1.039
1.057
1.009
1.009
1.010
NAVY
CASA
1.00
1.10
1.25
1.50
1.75
2.00
3.00
4.00
1.177
1.170
1.161
1.148
1.137
1.127
1.099
1.079
1.167
1.151
1.131
1.103
1.079
1.059
1.001
0.961
MSSP (R)
20.000
19.630
20.886
23.526
22.419
21.199
1.250
2.740
1.469
2.082
2.082
2.082
13.558
3.019
8.837
4.422
4.422
4.422
MLSP (L)
Life Factor (NF)
Since MLSP from the Liberty data (scenario 1) are lower than for the NAVY database, the life
factor is reduced to 8.837 from 13.558. This is further reduced to 4.422 after adding the FAALVM data. For a given MLSP, the life factor is fixed since it does not depend on MSSP.
Therefore, further improvements to MSSP results in a decrease in the slope of the curve, which
pivots about NF (LEF = 1), as shown in figure 55. Consequently, this lowers the LEF, especially
84
for smaller test durations. In scenario 3, MSSP is reduced when the Liberty adhesive data are
included, indicating a small increase in scatter for the Weibull distribution of static scatter
parameters. MSSP is further decreased in scenario 4, when element-level adhesive joint data
from FAA-EOD are included. FAA-EOD element test specimens are bonded using EA9394.
However, the Liberty XL2 is bonded using EPIBOND 1590. Therefore, scenario 3 is more
applicable to this structure.
1.30
NAVY
CASA
Scenario 1
Scenario 2
Scenario 3
Scenario 4
LEF
1.20
1.10
1.00
0.0
2.0
4.0
NF
6.0
8.0
10.0
Test Duration, N
1.30
LEF
1.20
1.10
NAVY
CASA
Scenario 1
Scenario 2
Scenario 3
Scenario 4
1.00
0.1
1.0
Test Duration, N
Further
Improvements
to MSSP
10.0
Figure 55. Influence of Test Duration on B-Basis LEFs for Liberty XL2
85
Although scenario 3 did not show significant improvements for LEF, the life factor was about
50 percent lower than for scenario 1. Also, B-basis LEFs for this scenario were considerably
lower than for NAVY. The original fatigue tests of the Liberty XL2 fuselage were conducted
using the NAVY life factor and an LEF of 1.15, which corresponds to a test duration of 1.5 DLT.
The Liberty fuselage fatigue test was conducted for three test lives with NAVY LEF and
included a damage tolerance phase. According to scenario 3, an LEF of 1.15 corresponds to a
test duration of 1 DLT. Thus, the full-scale test with an LEF of 1.15 corresponds to a B-basis
life factor of 1. The actual test was conducted with two DLTs more than what is required to
achieve the same level of reliability demonstrated by the metal structure. This increases the
confidence in DLT, i.e., number of hours corresponding to a test duration of 1 DLT.
During full-scale tests, certain spectrum load cycles above a percentage of the DLLs (high-load
cycles) were multiplied by the NAVY life factor rather than applying LEF, as described in
section 3.4. Due to the improvement in data scatter, the life factor was reduced to 4.422 from the
value of 13.558 proposed for composites by Whitehead, et al. [2]. Therefore, the high-load
cycles were repeated approximately three times more than the required number of repetitions to
achieve the B-basis life factor. This further increased the level of confidence in DLT.
To compare the B-basis LEF requirements for different databases, the Liberty XL2 fuselage
durability and damage tolerance (DaDT) test was compared against two scatter analysis material
databases: (1) NAVY and (2) Liberty data (scenario 4 in figure 55). This approach evaluated
the modified DLT in the event that a certification test duration or LEF was different from the
minimum required to achieve B-basis reliability. Figure 56 shows the B-basis LEF requirements
with respect to different test durations based on data from the above-mentioned two databases.
Liberty LEF requirements reflect the material and process as well as test method improvements
related to composites, which consequently reduced the data scatter compared to the test data used
for the NAVY approach. Points C and Z correspond to the life factor for Liberty and NAVY
databases, respectively. Points X and Y show the LEF required for 1.5- and 3-DLT test
durations, respectively, according to the NAVY approach. For the same test durations, points A
and B show the LEF requirements for the improved composite test data for Liberty XL2 fuselage
materials.
A full-scale DaDT test of the 1653-pound Liberty XL2 fuselage was performed at NIAR for
three lifetimes. Point T in figure 56 shows the LEF used during the DaDT full-scale test
substantiation. The load spectra used for cyclic loading was developed using the exceedance
curves from DOT/FAA/CT-91/20 [47] (maneuver and gusts) and AFS-120-73-2 [48] (taxi and
landing). The test spectra were truncated below the 30-percent DLL to shorten the test without
significant effects on the fatigue characteristics of the structure. No load in excess of DLL was
applied during cyclic tests.
Since Liberty LEF testing was in progress at the time of full-scale test substantiation, an LEF of
1.15 corresponding to 1.5 lifetimes per the NAVY approach [2] was applied to loads below an
80-percent limit load, and an NF of 28.5 (conservatively, this factor was selected during full-scale
substantiation until the LEF data is available) was applied for loads above an 80-percent limit
load, similar to the approach outlined in figure 19(b). This approach allowed for a lower LEF in
86
a trade-off for more life cycles, which would reduce the severity of the overload for high stress
levels.
1.30
1.20
T
LEF
X
Y
A
1.10
NAVY
Liberty
B
1.00
0.1
C
1.0
10.0
Z
100.0
Test Duration, N
Figure 56. Comparison of Tested and Required LEFs for Liberty XL2
Based on the NAVY approach, B-basis reliability was obtained either by increasing the loads by
an LEF of 1.15 for a test duration of 1.5 DLTs or by applying an NF of 13.6. This is equivalent
to an NF of 2 DLTs, which is typically used to demonstrate B-basis reliability of metal structures.
However, as shown in figure 56, the Liberty XL2 fatigue test article was tested for twice the
required test duration (3-DLTs) with no indication of damage growth, demonstrating residual
strength after repeated loading. Therefore, it is evident that the test article demonstrated B-basis
reliability for twice the DLT of 5000 hours. Typically, this argument alone would not justify the
new DLT, as the load segments above 80-percent DLL are multiplied by NF instead. However,
in this case, NF should have been 13.6 or 4.422, based on the NAVY or Liberty approach,
respectively. The applied NF is more than twice that of the NAVY approach and six times that of
the Liberty approach. Thus, the above-mentioned argument for a design lifetime of 10,000 hours
is justified. An NF of 28.5 corresponds to an MLSP, or αL of 1.00 (figure 14 or solving equation
8 for αL), which corresponds to materials that exhibit higher scatter in fatigue life data than the
data set in the NAVY approach.
Alternately, considering only the NAVY approach and based on the difference between applied
and required LEFs for B-basis reliability, it can be shown that this structure is capable of
carrying 5 percent more load than the tested load spectrum for substantiation of a 5000-hour
DLT. Based on the test data applicable to Liberty XL2 fuselage materials and design details, an
87
LEF of 1.035 for three lifetimes (table 40), or an NF of 4.422, was sufficient to demonstrate
B-basis reliability for a 5000-hour DLT. Therefore, considering the improved LEF curve and
based on the difference between applied and required LEFs for B-basis reliability, it can be
shown that a 1653-pound fuselage structure is capable of carrying 11.5 percent more loads than
the designed load spectrum for a DLT of 5000 hours. This additional information can be used to
support design changes that result in higher load requirements, given that the load spectra, failure
modes, and critical locations of the structure are not changed.
This case study addresses concerns related to the application of LEF and NF during full-scale test
substantiation. Once the LEF curve is generated, as shown in figure 56, any combination of LEF
and test duration, N, provides the same level of reliability that is required to be demonstrated by
metal structures, N = 2-DLT. Thus, the test duration and life factor are no longer required to be
multiplied by 2 to achieve B-basis reliability for the designed lifetime.
5.6 CASE STUDY—SCATTER ANALYSIS OF BONDED JOINTS.
To investigate the scatter in adhesive data with respect to composite data, a scatter analysis was
conducted on adhesive data generated on the composite-titanium bonded joint shown in figure
57. Berens and West [49] conducted static tests in different test temperatures and loading rates,
while conducting fatigue tests in two different temperatures and multiple stress levels. The
Weibull analysis for static test data and fatigue test data are shown in tables 41 and 42,
respectively. This data set is referred to as STP580.
Figure 57. Composite-Titanium Bonded Joint Test Specimen [49]
88
Table 41. Weibull Parameter Estimates for Static Tests [49]
Test Temperature
(°F)
-40
73
150
200
250
300
Loading Rate
(lb/min)
120
1200
12000
120
1200
12000
120
1200
12000
1200
120
1200
12000
120
1200
12000
Number of
Specimens
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
10
Scale
Parameter, β̂
(psi)
5165
5756
5638
5057
4955
5145
4449
4510
4624
3949
2840
3143
3486
1693
2148
2879
Shape
Parameter,
α̂
8.01
11.37
13.87
12.32
8.44
14.41
11.37
8.88
14.26
12.98
22.06
13.68
16.01
10.95
9.08
16.51
Table 42. Weibull Parameter Estimates for Fatigue Tests [49]
Test Temperature
(°F)
73
200
Maximum
Load (lb)
3100
2900
2700
2500
2350
2150
2000
2000
Number of
Specimens
10
10
10
9
6
8
9
15
Scale
Parameter, β̂
(Cycles)
53
140
124
1020
777
485
2870
367
Shape
Parameter,
α̂
2.62
2.22
1.99
1.43
0.92
1.56
0.69
1.34
Shape parameters corresponding to different static data sets in table 41 were pooled and the
shape and scale parameters that correspond to their probability density function were calculated
as 3.733 and 14.088, respectively. Then, the MSSP for STP580 data was calculated as 12.959.
89
Since there was not enough fatigue data, the shape parameters in table 42 were considered as
separate fatigue data sets and the shape and scale parameters corresponding to their probability
density function were calculated as 2.898 and 1.796, respectively. Then, the MLSP for STP580
data was calculated as 1.552. Note that, typically, shape parameters obtained for several S/N
curves are analyzed to obtain MLSP, as shown in the examples in this section. In the absence of
multiple S/N curves, the distribution of Weibull shape parameters obtained for multiple stress
levels can be used to generate MLSP as long as there are data from more than six stress levels
and each stress level has more than six fatigue (failure) data points. This method is introduced
through this case study and referred to as the modified individual Weibull analysis. Unlike the
individual Weibull method that calculates the arithmetic mean of shape parameters obtained for
each stress level, the modified individual Weibull method considers the Weibull distribution of
shape parameters in each stress level to calculate the MLSP that represents the data scatter of the
S/N curve. For the 73°F data in table 42, the MLSP was calculated as 1.578 using shape and
scale parameters corresponding to the modified individual Weibull method (2.836 and 1.839,
respectively). The individual Weibull analysis for this data set resulted in a shape parameter of
1.633. Therefore, the modified individual Weibull method is considered more conservative than
the individual Weibull method and is recommended for adhesive data that satisfy the abovementioned requirements.
To compare the STP580 fatigue scatter analysis data with some newer adhesive joint data, the
shape parameters corresponding to adhesive joint data were extracted from the FAA-LEF and
FAA-D5656 databases. The strength shape parameters of SLS data in the FAA-LEF database
(tables 8 and 11) were used for calculating the MSSP, while the FAA-D5656 data set was used
for calculating the MLSP. The scatter analysis of bonded joint static data in the FAA-LEF
database resulted in shape and scale parameters of 2.181 and 28.262, respectively. The MSSP
corresponding to these parameters was calculated as 21.334. The MLSP corresponding to newer
forms of bonded joints was selected as 1.461 from the individual Weibull analysis data shown in
table 34 using the MLE approach. These modal strength and life shape parameters are tabulated
in table 43 along with the NAVY and NIAR data (figure 52). Then, the LEFs obtained from
STP580 are compared with the LEFs obtained from the NAVY, NIAR, and FAA-LEF/FAAD5656 adhesive data sets (figure 58). With the limited amount of bonded joint data included in
these analyses, especially in STP580, it is evident that the fatigue life data scatter in bonded
joints is higher than the newer composite data. Note that except for NIAR data, the rest of the
fatigue scatter analyses were conducted using individual Weibull. The NIAR fatigue data shown
here was analyzed using the Sendeckyj method, which is significantly more conservative than
the individual Weibull method.
Table 43. Summary of Strength and Life Shape Parameters of Composite and Adhesive Data
MSSP
MLSP
NAVY
20.000
1.250
NIAR
26.310
2.131
STP580
12.959
1.552
90
FAA-LEF (Adhesive)
and FAA-D5656
21.334
1.461
1.200
NAVY
1.175
NIAR
STP580
FAA-LEF (Adhes ive) & FAA-D5656
1.150
LEF
1.125
1.100
1.075
1.050
1.025
1.000
0.0
2.0
4.0
6.0
8.0
Test Duration, N
10.0
12.0
14.0
Figure 58. Comparison of LEFs Generated Using Composite and Adhesive Test Data
Contrary to the observation in fatigue data scatter, table 43 shows a significant improvement in
the static data scatter in newer bonded joints compared to STP580 data, which is possibly due to
improvements in adhesives, processes, and test methods. Thus, the reliability of bonded joint
test data is still a concern for durability and damage tolerance test substantiation and must be
included in the development of LEFs, if bonded joints are present in a structure. Furthermore,
the use of individual Weibull analysis of bonded joint fatigue data is recommended for
preventing overly conservative LEFs as significant differences are observed on the shape
parameters obtained at different stress levels (table 42). An LEF curve based on newer bonded
joint test data indicates that the NAVY values are conservative.
6. CONCLUSIONS AND RECOMMENDATIONS.
The procedure for generating load enhancement factors (LEF) based on the most critical design
details of a composite structure at the coupon level was developed. The approach for obtaining
the modal static-strength shape parameter (MSSP) and modal fatigue-life shape parameter
(MLSP) to calculate the LEFs for different test durations was investigated in detail using static
and fatigue test data for several different composite material systems. It was shown that the
scatter in notched (damaged) composite test data was significantly lower than in unnotched
composite. Such improvements in fatigue-life shape parameter can significantly reduce the life
factor. However, the life factor becomes insensitive to small changes in the life shape parameter
beyond a value of 4, which is considered to be the life shape parameter for metal. The composite
MLSP of 1.25, which was used for the U.S. Navy F/A-18 certification (or NAVY) approach, lies
within the highly sensitive region of the life factor versus shape parameter curve; thus, even a
91
small improvement resulted in a dramatic reduction in life factor, which reflects the required
number of test durations to achieve a certain level of reliability in the design lifetime. The
analysis in this report forms the supporting data for the load-life damage hybrid approach that
can be applied to a full-scale durability test article during the certification process.
There are several approaches for the scatter analysis of fatigue data, including the individual
Weibull, joint Weibull, and Sendeckyj wearout models. When analyzing small fatigue data sets,
the latter two methods can be used to pool data across fatigue stress levels. Furthermore,
Sendeckyj analysis allows the user to include the static and residual strength of run-out
specimens. In addition to a probabilistic description of the data scatter, the Sendeckyj wearout
model provides a deterministic equation to define the shape of the stress to number of cycles
(S/N) curve and an expression for the monotonically decreasing residual strength as a function of
the number of cycles.
Compared to metals, composite materials are known for higher scatter in both static and fatigue
test data due to their heterogeneous nature, higher sensitivity to batch variability, environment,
and complex failure modes. Over the years, improvements in test methods, materials, and
process techniques have resulted in a significant reduction in data scatter. A detailed scatter
analysis conducted on several material test databases (from past and current Federal Aviation
Administration-funded research programs) representing multiple batches, loading modes,
environments, and laminate stacking sequences for several commonly used composite material
systems has shown that both static-strength and fatigue-life scatter have been reduced
significantly. These improvements have a direct impact on the probabilistic or reliability-based
analysis techniques for predicting the life of a composite structure, such as life factor and LEF
analysis.
It is recommended that specimens or elements representative of features of a particular structure
(i.e., materials, design details, failure modes, loading conditions, and environments) be included
in the analysis rather than pooling various material databases. Also, it is noted that the primary
goal in scatter analysis is not to select shape parameters from the critical lay-up, R-ratio,
environment, etc. (which may result in skewed data that will produce an unconservative LEF),
but rather to select the design details representing the critical areas of the structure. It is essential
to note that all critical design details and/or loading modes, i.e., open-hole fatigue with a stress
ratio of -1 for carbon composites, may result in low fatigue life due to notch effects and severe
load reversal. Such data often tend to produce low data scatter. Therefore, the details with poor
fatigue characteristics or fatigue critical designs may not always produce large data scatter and
may produce unconservative LEFs. When designing test matrices for generating fatigue shape
parameters, it is essential to investigate the design details/conditions that will produce the most
data scatter, which will consequently affect the reliability of test data in higher levels of building
blocks of testing.
It is important that the test matrix include sufficient information to translate the statistical
significance of such phenomenon in a meaningful manner into a full-scale test substantiation.
Such test matrices can be significantly reduced by focusing on critical aspects of the structure to
address the minimum requirements. As demonstrated in the case study of the Liberty XL2
fuselage, the use of shared databases can significantly reduce the amount of additional tests and
92
time required for a certain application, but care must be taken to make certain that the shared
data are equivalent to what is used for that application. Adhesive joints, if applicable, may
require the use of individual Weibull analysis rather than pooling techniques such as the
Sendeckyj analysis method, as adhesive joints tend to produce large scatter, mainly due to
imperfections during bonding and high sensitivity to load eccentricity (especially in asymmetric
joints). In the absence of multiple S/N curves, the distribution of Weibull shape parameters
obtained for multiple stress levels can be used to generate MLSP as long as there are data from
more than six stress levels and each stress level has more than six fatigue (failure) data points.
This method is introduced through the case study of bonded joint data and referred to as the
modified individual Weibull analysis. Unlike the individual Weibull method that calculates the
arithmetic mean of shape parameters obtained for each stress level, the modified individual
Weibull method considers the Weibull distribution of shape parameters (indicator of data scatter)
in each stress level to calculate shape parameter that represents the data scatter of the S/N curve.
The modified individual Weibull method is considered more conservative than the individual
Weibull method and recommended for adhesive data that satisfy the above-mentioned
requirements.
Although the shape parameters can vary within a large spectrum of values, care must be taken to
address unrealistic values individually to produce safe and reliable scatter analysis. For example,
the Sendeckyj model provides a way to graphically inspect the data fit, as illustrated in appendix
A, to evaluate the quality of the fitting parameters, which are used to generate the S/N curve fit
and the scatter analysis. For Weibull analysis, the maximum-likelihood estimation (MLE) can
be used to generate shape parameters when there are more than 18 data points, while regression
in the number of cycles is recommended for small data sets. For the MLSP calculations of
composite data in this report, both techniques produced similar values, while for adhesives,
regression in the number of cycles produced higher shape parameters than from the MLE.
7. REFERENCES.
1.
Aniversario, R.B., Harvey, S.T., McCarty, J.E., Parson, J.T., Peterson, D.C., Pritchett,
L.D., Wilson, D.R., and Wogulis, E.R., “Full-Scale Testing, Production and Cost
Analysis Data for the Advanced Composite Stabilizer for Boeing 737 Aircraft,” Vol. 2,
NASA CR-166012, December 1982.
2.
Whitehead, R.S., Kan, H.P., Cordero, R., and Saether, E.S., “Certification Testing
Methodology for Composite Structures,” Report No. NADC-87042-60, Volumes I and II,
October 1986.
3.
Lameris, J., “The Use of Load Enhancement Factors in the Certification of Composite
Aircraft Structures,” NLR Report: NLR TP 90068 U, February 1990.
4.
Seneviratne, W.P., “Fatigue-Life Determination of Damage-Tolerant Composite
Airframe,” Dissertation, Wichita State University, December 2008.
5.
Sanger, K.B., “Certification Testing Methodology for Composite Structures,” Report No.
NADC-86132-60, January 1986.
93
6.
Deo, R., Starnes, Jr., J.H., and Holzwarth, R., “Low-Cost Composite Materials and
Structures for Aircraft Applications,” NATO Applied Vehicle Technology Panel (AVT)
Specialists’ Meeting, RTO-MP-069(II), Leon, Norway, May 2001.
7.
Kan, H.P. and Whitehead, R.S., “Damage Tolerance Certification Methodology for
Composite Structures,” Proceedings of Eighth DoD/NASA/FAA Conference on Fibrous
Composites in Structural Design, NASA CP-3087, Part 2, Norfolk, Virginia, November
1989, pp. 479-498.
8.
Shah, C.H., Kan, H.P., and Mahler, M., “Certification Methodology for Stiffener
Termination,” FAA report DOT/FAA/AR-95/10, April 1996.
9.
Kan, H.P. and Dyer, T.A., “Structural Integrity Evaluation of the Lear Fan 2100
Aircraft,” FAA report DOT/FAA/AR-95/13, May 1996.
10.
McCarty, J.E., Johnson, R.W., and Wilson, D.R., “737 Graphite-Epoxy Horizontal
Stabilizer Certification,” Proceedings of the 23rd AIAA/ASME/ASCE/AHS Structures,
Structural Dynamics and Materials Conference, May 1982, pp. 307-322.
11.
Gogkol, O., “A310-300 CRFP Fin – Damage Tolerance Demonstration,” Proceedings of
the 7th International Conference of SAMPE, European Chapter, Munich, June 1986.
12.
Brandecker, B. and Hilgert, R., “A320 Full Scale Structural Testing for Fatigue and
Damage Tolerance Certification of Metallic and Composite Structure,” Proceedings
of the 16th Congress of the International Council of the Aeronautical Sciences,
ICAS-88-5.8.1, Jerusalem, Israel, 1988.
13.
Fawcett, A., Trostle, J., and Ward, S., “777 Empennage Certification Approach,”
Proceedings of the 11th International Conference on Composite Materials, Australia,
July 1997.
14.
Vosteen, L.F. and Hadcock, R.N., “Composite Chronicles: A Study of the Lessons
Learned in the Development, Production, and Service of Composite Structures,”
NASA CR-4620, November 1994.
15.
Wong, R. and Abbott, R., “Durability and Damage Tolerance of Graphite/Epoxy
Honeycomb Structures,” Proceedings of the 34th International Symposium, Anaheim,
California, 1990.
16.
Kan, H.P. and Kane, D.M., “Probabilistic Certification of Integrally Bonded Composite
Structures—an Assessment,” Proceedings of the 43rd AIAA/ASME/ASCE/AHS/ASC
Structures, Structural Dynamics and Materials, Denver, Colorado, April 2002.
17.
Sumich, M. and Kedward, K.T., “Development of a Fatigue-Life Methodology for
Composite Structures Subjected to Out-of-Plane Load Components,” NASA TM-102885,
February 1991.
94
18.
Halpin, J., Jerina, K., and Johnson, T., “Characterization of Composites for the Purpose
of Reliability Evaluation,” Analysis of Test Methods for High Modulus Fibers and
Composites, ASTM STP 521, American Society for Testing Materials, 1973, pp. 5-64.
19.
Sendeckyj, G.P., “Fitting Models to Composite Materials Fatigue Data,” Test Methods
and Design Allowables for Fibrous Composites, ASTM STP 734, Chamis, C.C., ed.,
American Society for Testing and Materials, 1981, pp. 245-260.
20.
Sendeckyj, G.P., “Life Prediction for Resin Matrix Composite Materials,” Fatigue of
Composite Materials, Reifsnider, K.L., ed., Elsevier, Amsterdam, 1990.
21.
O’Brien, T.K. and Reifsnider, K.L., “Fatigue Damage Evaluation Through Stiffness
Measurements in Boron-Epoxy Laminates,” Journal of Composite Materials, No. 15,
1981, pp. 55-70.
22.
Hahn, H.T. and Kim, R.Y., “Fatigue Behavior of Composite Laminates,” Journal of
Composite Materials, Vol. 10, No. 2, 1976, pp. 156-180.
23.
Hwang, W. and Han, K.S., “Cumulative Damage Model and Multi-Stress Fatigue Life
Prediction,” Journal of Composite Materials, Vol. 20, No. 2, 1986, pp. 125-153.
24.
Hwang, W. and Han, K.S., “Fatigue of Composites—Fatigue Modulus Concept and Life
Prediction,” Journal of Composite Materials, Vol. 20, No. 2, 1986, pp. 154-165.
25.
Mahfuz, H., Zaman, K., Haque, A., Foy, C., Mohamed, H., and Jeelani, S., “Fatigue Life
Prediction of Thick-Section S2-Glass/Vinyl-Ester Composites Under Flexural Loading,”
Journal of Engineering Materials and Technology, Vol. 122, No. 4, October 2000,
pp. 402-408
26.
Halpin, J.C., Jerina, K.L., and Johnson, T.A., “Characterization of Composites for the
Purpose of Reliability Evaluation,” AFML-TR-77-289, Wright-Patterson Air Force Base,
Ohio, 1972.
27.
Jeans, L.L., Grimes, G.C., and Kan, H.P., “Fatigue Spectrum Sensitivity Study for
Advanced Composite Materials,” U.S. Air Force Flight Dynamic Laboratory, Technical
Report AFWAL-TR-80-3130, Volumes I, II, and III, 1980.
28.
Rosenfeld, M.S. and Huang, S.L., “Fatigue Characteristics of Graphite/Epoxy Laminates
Under Compression Loading,” Journal of Aircraft, Vol. 15, No. 5, 1978, pp. 264-268.
29.
Agarwal, B.D. and James, W.D., “Prediction of Low-Cycle Fatigue Behavior of GFRP:
An Experimental Approach,” Journal of Materials Science, Vol. 10, No. 2, 1975,
pp. 193-199.
30.
Yang, J.N. and Du, S., “An Exploratory Study Into the Fatigue of Composites Under
Spectrum Loading,” Journal of Composite Materials, Vol. 17, No. 6, 1983, pp. 511-526.
95
31.
Kassapoglou, C., “Fatigue Life Prediction of Composite Structures Under Constant
Amplitude Loading,” Journal of Composite Materials, Vol. 41, No. 22, 2007.
32.
Tomblin, J.S. and Seneviratne, W.P., “FAA Research on Large-Scale Test
Substantiation,” FAA Damage Tolerance and Maintenance Workshop, Rosemont,
Illinois, July, 2006.
33.
Tomblin, J.S. and Seneviratne, W.P., “Laminate Statistical Allowable Generation for
Fiber-Reinforced Composite Materials: Lamina Variability Method,” FAA report
DOT/FAA/AR-06/53, January 2009.
34.
Shyprykevich, P., Seneviratne, W.P., and Tomblin, J.S., “Lamina Variability Method for
Determining Laminate Basis Values,” 22nd Annual Technical Conference of the
American Society for Composites (ASC), Seattle, Washington, September 2007.
35.
Tomblin, J.S., Seneviratne, W.P., Escobar, P., and Yap, Y., “Fatigue and Stress
Relaxation of Adhesives in Bonded Joints,” FAA report DOT/FAA/AR-03/56,
Ocrtober 2003.
36.
Tomblin, J.S., Seneviratne, W.P., and Pillai, G.R., “Effects of Disbonds, Lightning
Strikes, and Low-Velocity Impact Damage on Adhesively Bonded Composite Joints,”
FAA report DOT/FAA/AR-09/4, December 2009.
37.
Composite Material Handbook (CMH-17), Volume 2: Polymer Matrix Composites
Material Properties, Chapter 4—Carbon Fiber Properties, Material Sciences Corp., in
cooperation with ASTM, 1999.
38.
Tomblin, J., Harter, P., Seneviratne, W., and Yang, C., “Characterization of Bondline
Thickness Effects in Adhesive Joints,” ASTM Journal of Testing and Evaluation,
JCTRER, Vol. 24, No. 2, 2002, pp. 332-344.
39.
Tomblin, J.S., Seneviratne, W.P., Kim, H., and Lee, J., “Box Beam Lap Shear Torsion
Testing for Evaluating Structural Performance of Adhesive Bonded Joints,” Joining and
Repair of Composites Structures, ASTM STP 1455, Kedward, K.T. and Kim, H., eds.,
ASTM International, 2004, pp. 42-54.
40.
ARAMIS, v. 5.3.0 User Manual, GOM mbH.
41.
Tomblin, J.S., Seneviratne, W.P., and Borgman, M.D., “Electronic Speckle Pattern
Interferometry for Investigating Defect Propagation of Honeycomb Sandwich
Specimens,” SAMPE Technical Conference, Baltimore, Maryland, November 2001.
42.
Life Data Analysis Reference, ReliaSoft Corporation, Tucson, Arizona.
43.
Badaliance, R. and Dill, H.D., “Compression Fatigue Life Prediction Methodology for
Composite Structures,” NADC-83060-60, Volumes I and II, September 1982.
96
44.
Rust, S.W., Todt, F.R., Harris, B., Neal, D., and Vangel, M., “Statistical Methods for
Calculating Material Allowables for MIL-HDBK-17,” Test Methods for Design
Allowables for Fibrous Composites, ASTM STP 1003, Chamis, C.C. ed., ASTM,
Philadelphia, Pennsylvania, 1989, pp. 136-149.
45.
Whitehead, R.S. and Schwarz, M.G., “The Role of Fatigue Scatter in the Certification of
Composite Structures,” ASTM Symposium on the Long-Term Behavior of Composites,
Williamsburg, Virginia, March 1982.
46.
Seneviratne, W.P., “Fatigue Analysis of Liberty XL2 Fuselage Materials,” National
Institute for Aviation Research, STR-RP-2008-005, September 2008.
47.
Gabriel, E.A., DeFiore, T., Locke, J.E., and Smith, H.W., “General Aviation Aircraft—
Normal Acceleration Data Analysis and Collection Project,” FAA report DOT/FAA/CT91/20, February 1993.
48.
“Fatigue Evaluation of Wing and Associated Structure of Small Airplanes,” FAA report
AFS-120-73-2, May 1973.
49.
Berens, A.P. and West, B.S., “Evaluation of an Accelerated Characterization Technique
for Reliability Assessment of Adhesive Joints,” Composite Reliability, ASTM STP 580,
ASTM, Philadelphia, Pennsylvania, 1975, pp. 90-101.
97/98
APPENDIX A—FATIGUE TEST RESULTS OF FEDERAL AVIATION ADMINISTRATION
LOAD-ENHANCEMENT FACTOR DATABASE
This appendix contains the stress to number of cycles (S/N) curves that were used for the fatigue
life scatter analysis of the Federal Aviation Administration load-enhancement factor (FAA-LEF)
database. Sendeckyj analysis [A-1] was conducted on the S/N data, and the fitting curves are
displayed here for a graphical confirmation that the analysis represents a reasonable trend. S/N
curves based on the Sendeckyj analysis are compared with the life predictions based on the
Kassapoglou method [A-2], which only uses static-strength data to predict fatigue life. In
addition, for several selected fatigue specimens, the compliance change and the damage growth
are compared.
A.1 THE S/N DATA FOR AS4/E7K8 PLAIN-WEAVE FABRIC.
This section contains the S/N data for the AS4/E7K8 plain-weave fabric tests included in the
FAA-LEF database. Tables A-1 through A-6 include the individual data points, and figures A-1
through A-14 show the S/N curves that were used for generating LEFs for AS4-PW. Figure A15 shows the Goodman diagram for AS4-PW open-hole (OH) test data. In these tables, n is the
number of cycles survived and n = 1 indicates static failure. Also, A and R correspond to the
fatigue stress level (or static failure stress level) and the residual strength after surviving the
corresponding number of cycles, respectively.
A-1
Table A-1. S/N Data for AS4-PW 10/80/10 OH Tests (FAA-LEF)
OH Compression/Tension
(R = -1)
 A
41228
39404
40497
39811
43154
38740
30354
26307
26307
26307
26307
26307
26307
26307
26307
20236
20236
20236
20236
20236
20236
20236
20236
16189
1
1
1
1
1
1
301
2195
1407
1412
1751
1996
1442
1927
4746
36171
29470
31608
32681
30972
26187
30657
75965
549419
 A
41228
39404
40497
39811
43154
38740
30354
30354
30354
30354
30354
30354
30354
28330
28330
28330
28330
28330
28330
28330
28330
26307
25497
25497
16189
652440
16189
16189
16189
16189
16189
16189
545138
715247
519140
503585
881812
771513
N
 R
OH Compression
(R = 5)
OH Tension
(R = 0)
1
1
1
1
1
1
2645
15660
11740
9151
10990
8239
11057
69069
44082
98781
90522
114108
52521
50311
70955
188105
229685
445665
 A
43792
44405
43580
43112
42126
43717
34764
34764
34764
34764
34764
34764
32591
32591
32591
32591
32591
32591
28246
28246
28246
28246
28246
28246
1
1
1
1
1
1
20505
15422
10607
11684
6077
11195
38373
55456
46146
71250
57471
54131
474638
377554
368844
314495
365748
389959
25497
348791
21728
1000000
25497
25497
25497
24283
24283
443210
726570
574103
1000000
1000000
N
 R
36385
34324
A-2
N
OH Tension
(R = -0.2)
 R
41546
 A
43792
44405
43580
43112
42126
43717
32591
32591
32591
32591
32591
32591
28246
28246
28246
28246
28246
28246
26073
26073
26073
26073
26073
26073
1
1
1
1
1
1
18137
18575
21301
22457
34293
17588
153000
119454
31998
151318
142394
178984
226885
390390
451383
270902
425390
281893
26073
332591
23900
23900
23900
1000000
1000000
1000000
N
 R
30517
29613
26546
Table A-2. S/N Data for AS4-PW 10/80/10 Compression After Impact and Double-Notched
Compression Tests (FAA-LEF)
Compression After Impact
(R = 5) [20 ply] Barely Visible Impact Damage
Compression After Impact
(R = 5) [40 ply] Visible Impact Damage
 A
34974
35928
32913
27684
27684
27684
27684
27684
27684
25954
25954
25954
25954
25954
25954
24224
22493
22493
22493
22493
22493
22493
20763
 A
28945
31307
30476
30525
29743
30585
22698
22698
22698
22698
22698
22698
21184
21184
21184
21184
21184
19671
19671
19671
19671
19671
n
1
1
1
9071
5856
8980
16161
10644
9777
34539
66766
42237
17223
43665
46917
99627
454828
740070
650366
468695
450007
709191
1000000
 R
n
 R
1
1
1
1
1
1
9471
15663
7994
21448
6833
8538
29471
47593
28418
63444
46077
601081
203021
145252
538785
374069
35656
A-3
Double-Notched
Compression
(R = -1)
 A
4236
3804
3836
3948
4037
4068
1595
1595
1595
1595
1595
1595
1595
1196
1196
1196
1196
1196
1196
997
997
997
997
997
997
678
n
1
1
1
1
1
1
7231
7524
6586
8621
8212
7256
7150
25573
35487
34290
47904
48215
59366
262210
361803
271419
307399
194263
104738
1000000
Double-Notched
Compression
(R = -0.2)
 R
3591
 A
4236
3804
3836
3948
4037
4068
1994
1994
1795
1795
1795
1795
1795
1795
1795
1595
1595
1595
1595
1595
1595
1595
1396
1396
1396
1396
1396
1396
1196
1196
1196
n
 R
1
1
1
1
1
1
25321
28298
27417
17379
10624
28230
54216
27694
16090
72855
99058
50752
39812
40000
34484
170739
393302
238336
265252
170266
221148
121619
711710
1000000
1000000
3381
3540
Table A-3. S/N Data for AS4-PW 0/100/0 Compression After Impact and Double-Notched
Compression Tests (FAA-LEF)
OH Tension
(R = -1)
Tension After Impact (R = 0) Visible Impact Damage
 A
17631
17409
17550
18326
18078
 A
21231
21970
22082
21821
22254
21890
13125
1
1
1
1
1
1
1211
14239
137
 A
15690
15744
14612
14150
15333
15180
10583
10937
6730
9789
500
9071
860
10937
10937
13547
7729
9789
8899
392
3001
9071
9071
808
976
10937
7957
8899
2709
9071
1933
10937
10937
8750
8750
8750
8750
8750
8750
7656
7656
7656
7656
7656
7656
6562
7893
6812
57561
83056
47243
65033
121089
66003
210460
191852
194105
216642
216727
254909
990178
8899
8899
8899
8899
8899
8009
8009
8009
8009
8009
8009
7120
7120
7120
7120
7120
7120
6764
6764
6230
695
675
717
621
745
7164
9277
2982
7037
7196
5607
249488
448722
79665
138485
173554
153293
807275
840693
1000027
9071
9071
8315
8315
7559
7559
7559
7559
7559
7559
6803
6803
6803
6803
6803
6803
6501
6501
6047
6047
6047
667
768
1226
5471
4243
15903
5396
7275
7844
29967
66940
38624
70854
142973
107914
25250
556214
1000020
698405
1000011
1000022
n
 R
Tension After Impact (R = 0) Barely Visible Impact Damage
 R
n
1
1
1
1
1
A-4
12663
 R
n
1
1
1
1
1
1
464
13543
14565
Table A-4. S/N Data for AS4-PW 25/50/25 OH Tension and Compression After Impact Tests
(FAA-LEF)
OH Tension
(R = -1)
 A
44593
46643
43391
44080
45918
47623
27225
27225
27225
27225
27225
27225
22687
22687
22687
22687
22687
22687
20419
20419
20419
20419
20419
20419
n
1
1
1
1
1
1
14982
11988
15400
7335
8149
16101
129345
105310
142170
103758
117594
117183
446962
524270
604378
498321
949760
916940
Compression After
Impact (R = 5) Barely Visible Impact
Damage
Compression After
Impact (R = 5) Visible Impact Damage
Compression After
Impact (R = 5) Large Impact Damage
 A
37188
34745
35658
36526
36364
35669
28820
28820
28820
28820
28820
28820
27019
27019
27019
27019
27019
27019
25217
25217
25217
25217
25217
25217
 A
29149
31335
29443
29282
29950
28866
22374
22374
22374
22374
22374
22374
20882
20882
20882
20882
20882
20882
19391
19391
19391
19391
19391
19391
 A
25147
25601
24627
25370
25228
26695
19083
19083
19083
19083
19083
19083
16539
16539
16539
16539
16539
16539
15267
15267
15267
15267
15267
15267
n
1
1
1
1
1
1
12243
14342
9651
8152
15155
26005
92926
31634
104891
152023
47635
31642
678421
596825
323026
252255
575983
252433
A-5
n
1
1
1
1
1
1
37690
24001
55768
28958
11897
16335
127451
94625
128689
59749
143030
180742
626039
397153
270784
638545
222775
595875
n
1
1
1
1
1
1
42897
38476
18155
13719
32463
17564
201380
214807
374375
278234
165086
193821
2233805
1352887
1618147
1236307
928401
1228113
Table A-5. S/N Data for AS4-PW 40/20/40
Compression After Impact Tests
(FAA-LEF)
Table A-6. S/N Data for AS4-PW Sandwich
Tests (FAA-LEF)
Flexure (R = 0)
Compression After Impact
(R = 5) - Visible Impact Damage
 A
 A
143.828
145.592
146.042
146.530
147.151
139.028
87.000
87.000
87.000
87.000
87.000
87.000
87.000
72.500
72.500
72.500
72.500
72.500
72.500
72.500
65.250
58.000
58.000
58.000
58.000
58.000
58.000
58.000
 R
n
30623
1
31444
33538
31473
32526
31465
25476
25476
25476
25476
25476
23884
23884
23884
23884
23884
23884
20699
20699
20699
20699
20699
20699
20699
20699
20699
1
1
1
1
1
11230
21414
10473
9354
10449
37414
27422
31761
40216
59635
45263
538811
800295
849092
774653
860179
726956
1000030
1000032
1000051
31056
31272
30183
A-6
n
 R
1
1
1
1
1
1
26661
26077
23272
20000
22000
18898
19928
48648
170000
60000
80000
145000
190000
235000
150000
470000
580000
340000
500000
250000
420000
1000000
145.545
50000
45000
40000
35000
Kas s apoglou
Stress (psi)
30000
25000
Se nde ck yj
20000
Experiment
15000
Residual Strength
Equivalent Static Strength
10000
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
100,000
1,000,000
Cycles
Figure A-1. AS4/E7K8 PW—10/80/10, OH, R = -1
50000
45000
40000
35000
Kas s apoglou
Se nde ck yj
Stress (psi)
30000
25000
20000
Experiment
15000
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
Cycles
Figure A-2. AS4/E7K8 PW—10/80/10, OH, R = 5
A-7
50000
45000
Se nde ck yj
40000
Kas s apoglou
35000
Stress (psi)
30000
25000
20000
Experiment
15000
Residual Strength
Equivalent Static Strength
10000
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
100,000
1,000,000
Cycles
Figure A-3. AS4/E7K8 PW—10/80/10, OH, R = 0
50000
45000
Se nde ck yj
40000
35000
Kas s apoglou
Stress (psi)
30000
25000
20000
15000
Experiment
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
Cycles
Figure A-4. AS4/E7K8 PW—10/80/10, OH, R = -0.2
A-8
40000
35000
Se nde ck yj
30000
Stress (psi)
25000
Kas s apoglou
20000
15000
Experiment
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-5. AS4/E7K8 PW—10/80/10, CAI (20 ply)—BVID, R = 5
35000
30000
Kas s apoglou
25000
Stress (psi)
Se nde ck yj
20000
15000
Experiment
10000
Residual Strength
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-6. AS4/E7K8 PW—10/80/10, CAI (40 ply)—BVID, R = 5
A-9
1,000,000
4500
4000
3500
Kas s apoglou
Stress (psi)
3000
2500
2000
Se nde ck yj
1500
Experiment
Residual Strength
1000
Equivalent Static Strength
Sendeckyj
500
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-7. AS4/E7K8 PW—10/80/10, DNC, R = -1
5000
4500
4000
3500
Kas s apoglou
Stress (psi)
3000
2500
Se nde ck yj
2000
Experiment
1500
Residual Strength
1000
Equivalent Static Strength
Sendeckyj
500
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-8. AS4/E7K8 PW—10/80/10, DNC, R = -0.2
A-10
1,000,000
25000
Kas s apoglou
20000
Stress (psi)
15000
Se nde ck yj
10000
Experiment
Residual Strength
5000
Equivalent Static Strength
Sendeckyj
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-9. AS4/E7K8 PW—0/100/0, OH, R = -1
25000
20000
Kas s apoglou
Stress (psi)
15000
10000
Se nde ck yj
Experiment
Residual Strength
5000
Equivalent Static Strength
Sendeckyj
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-10. AS4/E7K8 PW—0/100/0, TAI—BVID, R = 0
A-11
1,000,000
18000
16000
14000
Kas s apoglou
Stress (psi)
12000
10000
Se nde ck yj
8000
6000
Experiment
Residual Strength
4000
Equivalent Static Strength
Sendeckyj
2000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-11. AS4/E7K8 PW—0/100/0, TAI—VID, R = 0
160
140
Kas s apoglou
120
Se nde ck yj
Stress (psi)
100
80
60
Experiment
Residual Strength
40
Equivalent Static Strength
Sendeckyj
20
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-12. AS4/E7K8 PW—HRH 10, Flexture, R = 0
A-12
1,000,000
50000
45000
40000
Kas s apoglou
35000
Se nde ck yj
Stress (psi)
30000
25000
20000
Experiment
15000
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-13. AS4/E7K8 PW—25/50/25, OH, R = -1
35000
30000
Kas s apoglou
Se nde ck yj
Stress (psi)
25000
20000
15000
Experiment
10000
Residual Strength
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-14. AS4/E7K8 PW—40/20/40, CAI—VID, R = 5
A-13
1,000,000
Normalized Alternating Stress
1.0
R= -1
R= -0.2
R= 0
0.8
R= 5
0
10 Cycles
3
0.6
10
5
10
0.4
6
10
0.2
0.0
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
Normalized Mean Stress
Figure A-15. Goodman Diagram Based on AS4/E7K8 PW OH Test Data
A.2 PROGRESSIVE DAMAGE FAILURE AND COMPLIANCE CHANGE.
The damage progression of several selected fatigue specimens was monitored by throughtransmission ultrasonic (TTU) C-scanning. The damage growth was then compared with the
compliance change of those specimens during fatigue tests. Compliance was measured by the
following:




Stopping the fatigue test and periodically conducting a quasi-static test (static)
Collecting load-displacement data during fatigue at several fatigue intervals (dynamic)
Using extensometer data (ext.)
Using a laser extensometer (laser)
These data are shown in figures A-16 though A-22 for several open-hole compression (OHC)
and flexure specimens.
A-14
Figure A-16. AS4/E7K8 PW—10/80/10, OHC, R = 5, Stress Level = 63% of Static
1.4
149000
146000
143000
1.0
Specimen Name: A117-BX25-A-6
SL : 75%
Max: -6.071 ksi (-1394 lbf)
Min: -30.354 ks i(-6968 lbf)
R: 5
Transducer : 1 mHz
F : 5 Hz
Gain : 82-90
0.8
0.6
140000
137000
134000
Slope (lbf/in)
Damaged Area (sq.in)
1.2
0.4
131000
Damaged Area [in2]
Slope (Dynamic)
Slope (Static)
0.2
128000
0.0
125000
0
2000
4000
6000
8000
No. of Cycles (n)
Figure A-17. AS4/E7K8 PW 01—10/80/10, OHC, R = 5, Stress Level = 75% of Static
(Specimen A)
A-15
150000
4.0
148000
3.5
144000
Damaged Area [in2]
Slope (Laser)
Slope (Ext.)
Slope (Dynamic)
Slope (Static)
2.5
142000
2.0
140000
1.5
Specimen Name: A117-BX25-A-7
SL : 75%
Max: -6.071 ksi (-1382 lbf)
Min: -30.354 ksi (-6909 lbf)
R:5
F : 5 Hz
Transducer : 1 mHz
Cycle : 11057
Gain : 82-90
1.0
0.5
138000
136000
134000
0.0
0
2000
4000
Slope (lbf/in)
Damaged Area (sq.in)
146000
3.0
6000
8000
10000
132000
12000
No. of Cycles (n)
4.0
160000
3.5
155000
3.0
150000
2.5
145000
2.0
140000
1.5
135000
Specimen Name: A117-BX35-A-10
Damaged Area [in2]
SL : 70%
Slope (Laser)
Max: -5.666 ksi (-1300 lbf)
Slope (Ext.)
Min: -28.330 ksi (-6501 lbf)
Slope (Dynamic)
R:5
Slope (Static)
F : 5 Hz
Transducer : 1 mHz
Cycle : 70955
Gain : 82-90
1.0
0.5
130000
125000
0.0
0
10000
20000
30000
40000
50000
Slope (lbf/in)
Damaged Area (sq.in)
Figure A-18. AS4/E7K8 PW—10/80/10, OHC, R = 5, Stress Level = 75% of Static
(Specimen B)
60000
70000
120000
80000
No. of Cycles (n)
Figure A-19. AS4/E7K8 PW—10/80/10, OHC, R = 5, Stress Level = 70% of Static
(Specimen A)
A-16
1.2
143000
138000
0.8
133000
0.6
128000
0.4
Specimen Name: A119-BX35-A-9
SL : 70%
Max: -5.666 ksi (-1303 lbf)
Min: -28.330 ksi (-6515 lbf)
R: 5
Transducer : 1 mHz
F : 5 Hz
Gain : 82-90
0.2
Damaged Area [in2]
Slope (Dynamic)
Slope (Static)
123000
0.0
0
10000
20000
30000
Slope (lbf/in)
Damaged Area (sq.in)
1.0
40000
50000
118000
60000
No. of Cycles (n)
Figure A-20. AS4/E7K8 PW—10/80/10, OHC, R = 5, Stress Level = 70% of Static
(Specimen B)
3000
Dynamic Slope (lbf/in)
2500
Specimen Name: A103-DX10-A-5
SL : 60%
Max: 0.00 ksi (0 lbf)
Min: 0.087 ksi (140 lbf)
R:0
F : 5 Hz
Cycle : 57249
Transducer : 1 mHz
Gain : 78- 83
Oscilloscope Sonic Workstation
2000
1500
1000
500
Slope (Dynamic)
Slope (Static)
0
0
8000
16000
24000
32000
40000
48000
56000
No. of Cycles (n)
Figure A-21. AS4/E7K8 PW and HRH10—Sandwich, Flexure, R = 0,
Stress Level = 60% of Static
A-17
3000
Dynamic Slope (lbf/in)
2500
Specimen Name: A103-DX20-A-3
SL : 50%
Max: 0.00 ksi (0 lbf)
Min: 0.073 ksi (116 lbf)
R:0
F : 5 Hz
Cycle : 122685
Transducer : 1 mHz
Gain : 78- 83
2000
1500
1000
500
Slope (Dynamic)
Slope (Static)
0
0
20000
40000
60000
80000
100000
120000
No. of Cycles (n)
Figure A-22. AS4/E7K8 PW and HRH10—Sandwich, Flexure, R = 0,
Stress Level = 50% of Static
Figures A-21 and A-22 show that the sandwich specimens carried the fatigue loads after
significant damage propagation because the sandwich facesheets carried most of the loads in
in-plane tension after the core shear capabilities were diminished. A 10-percent decrease in
compliance was considered as fatigue failure.
The compliance changes and the corresponding C-scan damage area for the three stress levels,
i.e., 75, 70, and 63 percent of OHC static strength, of several selected fatigue specimens with a
stress ratio of 5 are superimposed for comparison in figure A-23.
A-18
155000
145000
Higher stress level
125000
115000
105000
95000
85000
75000
1
10
100
1000
No. of Cycles [n]
10000
100000
1000000
(a) Compliance Change
0.80
0.70
0.60
2
Damaged Area (in )
Slope [lbf/in]
135000
0.50
0.40
A119-BX35-A-8
Higher stress level
SL1
SL2
0.30
SL3
0.20
0.10
0.00
1
10
100
1000
10000
100000
1000000
Cycles
(b) Damage Growth
Figure A-23. General Trends of Compliance Change and Damage Growth for
AS4-PW OH Fatigue Specimens
A-19
A.3 S/N Data for T700/#2510 Plain-Weave Fabric.
This section contains the S/N data for the T700/#2510 plain-weave fabric (T700-PW) tests
included in the FAA-LEF database. Tables A-7 and A-8 include the individual data points, and
figures A-24 through A-27 show the S/N curves that were used for generating LEFs for T700PW. In these tables, n is the number of cycles survived and n = 1 indicates static failure. Also,
A and R correspond to the fatigue stress level (or static failure stress level) and the residual
strength after surviving the corresponding number of cycles, respectively.
Table A-7. S/N Data for T700-PW OH Tests (FAA-LEF)
OHC/T (R = -1)
 A
33650
34576
35063
36487
35204
34936
24490
24490
24490
24490
24490
24490
20992
20992
20992
20992
20992
20992
19242
19242
19242
19242
19242
19242
17493
17493
17493
n
1
1
1
1
1
1
2532
2701
3810
3644
5948
6102
47865
32017
34468
54001
32295
38393
231143
251069
245405
112840
116223
146589
1056026
1046017
1000019
OHC (R = 5)
 R
29027
29668
27191
 A
33650
34576
35063
36487
35204
34936
31487
31487
31487
31487
31487
31487
29738
29738
27989
27989
27989
27989
27989
27989
26239
26239
26239
26239
26239
26239
26239
24490
n
1
1
1
1
1
1
2185
1889
1426
4263
2152
2466
27956
15207
89462
27907
38103
101532
16926
72487
455423
853210
427342
351708
218416
614718
1000014
1032657
OHT (R = 0)
 R
34559
35164
 A
41414
43186
41602
40934
42189
40789
35433
35433
35433
35433
35433
34599
34599
34599
34599
34599
33349
33349
33349
33349
33349
33349
31264
31264
31264
31264
31264
31264
30014
29180
29180
A-20
n
1
1
1
1
1
1
2338
780
500
331
1521
1879
3919
6211
12937
17960
21575
8256
17272
13863
18214
22073
169971
128793
188066
65969
159350
82926
240064
1089903
1000000
OHT (R = -0.2)
 R
35566
 A
41414
43186
41602
40934
42189
40789
33349
33349
33349
33349
33349
33349
31264
31264
31264
31264
31264
31264
29180
29180
29180
29180
29180
29180
29180
n
 R
1
1
1
1
1
1
8451
7003
8196
7766
11553
17718
36959
28014
43575
69444
71510
34351
157157
270388
524950
359994
381294
527262
1000183
29513
Table A-8. S/N Data for T700-PW DNC and Sandwich Tests (FAA-LEF)
DNC (R = -1)
 A
2878
3244
2786
3245
3009
2878
1503
1503
1503
1503
1503
1503
1353
1203
1203
1203
1203
1203
1203
1203
1052
1052
1052
1052
1052
1052
1052
1052
1052
1052
902
n
1
1
1
1
1
1
5331
7402
7835
1512
1661
11515
10811
51912
54388
44383
53626
174360
51686
43064
293357
243965
94645
207966
46212
291053
74427
32596
132350
144424
1000029
DNC (R = -0.2)
 R
 A
2878
3244
2786
3245
3009
2878
1804
1804
1804
1804
1804
1804
1503
1503
1503
1503
1503
1503
1503
1353
1353
1353
1353
1353
1353
1203
1203
1203
n
1
1
1
1
1
1
6616
5045
7781
4842
1757
6017
26712
40696
55920
38398
209297
40675
76358
403412
361713
226962
115567
150687
347703
473922
981488
1000023
2190
A-21
Flexure (R = 0)
 R
 A
131
137
139
138
138
141
82
82
82
82
82
82
82
82
82
69
69
69
69
69
69
62
62
62
62
62
62
62
n
1
1
1
1
1
1
11500
5100
5500
21000
22500
6750
24673
34800
24006
44036
60000
114000
78025
140000
123500
310133
110000
370000
340000
375000
492500
364181
 R
40000
35000
Kas s apoglou
30000
Stress (psi)
25000
Se nde ck yj
20000
15000
Experiment
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-24. T700-PW—10/80/10, OH, R = -1
40000
35000
Kas s apoglou
30000
Se nde ck yj
Stress (psi)
25000
20000
15000
Experiment
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
Cycles
Figure A-25. T700-PW—10/80/10, OH, R = 5
A-22
100,000
1,000,000
50000
45000
40000
Kas s apoglou
35000
Se nde ck yj
Stress (psi)
30000
25000
20000
Experiment
15000
Residual Strength
Equivalent Static Strength
10000
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-26. T700-PW—10/80/10, OH, R = 0
50000
45000
40000
Kas s apoglou
35000
Se nde ck yj
Stress (psi)
30000
25000
20000
Experiment
15000
Residual Strength
Equivalent Static Strength
10000
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
Cycles
Figure A-27. T700-PW—10/80/10, OH, R = -0.2
A-23
100,000
1,000,000
A.4 S/N DATA FOR 7781/#2510 8-HARNESS SATIN-WEAVE FABRIC.
This section contains the S/N data for the 7781/#2510 8-harness satin-weave fabric (7781-8HS)
tests included in the FAA-LEF database. Tables A-9 and A-10 include the individual data
points, and figures A-28 through A-31 show the S/N curves that were used for generating LEFs
for 7781-8HS. In these tables, n is the number of cycles survived and n = 1 indicates static
failure. Also, A and R correspond to the fatigue stress level (or static failure stress level) and
the residual strength after surviving the corresponding number of cycles, respectively.
Table A-9. S/N Data for 7781-8HS OH Tests (FAA-LEF)
OHC/T (R = -1)
 A
33130
33141
33506
33514
33407
32666
23256
16621
13297
13292
13292
10723
10723
10723
10723
10723
10723
10723
10013
9973
9973
9973
9959
9383
9383
9383
9383
9383
9383
8043
n
1
1
1
1
1
1
260
756
7502
4575
5335
32634
32224
33829
40038
30692
38903
40451
74386
81665
70390
362895
111843
123527
145031
134683
165288
179403
193388
593003
OHC (R = 5)
 R
 A
33130
33141
33506
33514
33407
32666
26582
26582
26582
26582
26582
26582
24921
24921
24921
24921
24921
24921
24921
24921
24921
23259
23259
23259
23259
23259
23259
22595
21598
19936
n
1
1
1
1
1
1
2813
2254
1587
1716
1673
1949
9408
29368
16482
27249
8833
13012
23041
7358
8213
216131
274091
595433
189941
347856
362656
1106426
1000000
1000100
OHT (R = 0)
 R
 A
27267
27150
26895
26613
26294
26632
16085
16085
16085
16085
16085
16085
13404
13404
13404
13404
13404
13404
10723
10723
10723
10723
10723
10723
29906
31661
30532
A-24
n
1
1
1
1
1
1
4879
4689
5153
4695
4478
5230
31239
29110
24024
23870
25278
29622
377715
352927
272790
353214
325874
288778
OHT (R = -0.2)
 R
 A
27267
27150
26895
26613
26294
26632
16085
16085
16085
16085
16085
16085
13404
13404
13404
13404
13404
13404
10723
10723
10723
10723
10723
10723
n
1
1
1
1
1
1
4159
3290
4340
4006
3901
5410
19991
26597
25380
24603
24317
22742
87248
193468
227929
193446
177910
193217
 R
Table A-9. S/N Data for 7781-8HS OH Tests (FAA-LEF) (Continued)
OHC/T (R = -1)
 A
8043
8043
8043
8043
8043
8043
OHC (R = 5)
n
 R
1000032
883245
942271
769637
778751
1144259
25937
 A
n
OHT (R = 0)
 R
 A
n
OHT (R = -0.2)
 R
 A
n
 R
Table A-10. S/N Data for 7781-8HS DNC and Sandwich Tests (FAA-LEF)
DNC (R = -1)
A
DNC (R = -0.2)
R
n
A
Flexure (R = 0)
R
n
A
R
n
3630
1
3630
1
140.342
1
3625
1
3625
1
141.508
1
3429
1
3429
1
141.772
1
3195
1
3195
1
143.715
1
3614
1
3614
1
139.256
1
3468
1
3468
1
139.038
1
1747
3121
2446
1692
83
28924
1747
5270
2096
9267
83
42000
1747
2028
2096
12301
83
39000
1747
6133
2096
18245
83
50000
1747
2759
2096
5612
83
36500
1747
3916
2096
15366
83
64000
1572
4004
2096
11326
70
265000
1572
10732
1747
65045
70
216004
1572
6779
1747
45870
70
205000
1397
28078
1747
99557
70
230000
1397
18684
1747
85112
70
167500
1397
5724
1747
105136
70
262000
1397
2138
1747
83695
70
70000
1397
6291
1572
320302
63
410000
1397
23500
1572
37069
63
270000
1223
451048
1572
455873
63
190000
1223
47332
1572
237356
60
325000
1223
31537
1572
281584
60
490000
1048
146896
1572
107805
60
625000
1048
206313
1572
168387
1048
379484
1397
1000088
1048
157833
60
790000
1048
197364
60
1000042
A-25
2745
60
1000042
60
1000042
Table A-10. S/N Data for 7781-8HS DNC and Sandwich Tests (FAA-LEF) (Continued)
DNC (R = -1)
A
R
n
1048
DNC (R = -0.2)
A
n
Flexure (R = 0)
R
57311
A
n
60
1000042
60
1000042
60
1000042
56
940000
40000
35000
Kas s apoglou
30000
Stress (psi)
25000
20000
Se nde ck yj
15000
Experiment
Residual Strength
10000
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-28. 7781-8HS—10/80/10, OH, R = -1
40000
35000
Kas s apoglou
30000
Se nde ck yj
Stress (psi)
25000
20000
15000
Experiment
Residual Strength
10000
Equivalent Static Sterngth
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-29. 7781-8HS—10/80/10, OH, R = 5
A-26
1,000,000
R
30000
Kas s apoglou
25000
Stress (psi)
20000
Se nde ck yj
15000
10000
Experiment
Residual Strength
Equivalent Static Strength
5000
Sendeckyj
Kassapoglou
0
1
10
100
1,000
10,000
100,000
1,000,000
Cycles
Figure A-30. 7781-8HS—10/80/10, OH, R = 0
35000
30000
Kas s apoglou
Stress (psi)
25000
20000
Se nde ck yj
15000
Experiment
10000
Residual Strength
Equivalent Static Strength
Sendeckyj
5000
Kassapoglou
0
1
10
100
1,000
10,000
100,000
Cycles
Figure A-31. 7781-8HS—10/80/10, OH, R = -0.2
A-27
1,000,000
A.5 REFERENCES.
A-1.
Sendeckyj, G.P., “Fitting Models to Composite Materials Fatigue Data,” Test Methods
and Design Allowables for Fibrous Composites, ASTM STP 734, Chamis, C.C., ed.,
ASTM, 1981, pp. 245-260.
A-2.
Kassapoglou, C., “Fatigue Life Prediction of Composite Structures Under Constant
Amplitude Loading,” Journal of Composite Materials, Vol. 41, No. 22, 2007.
A-28
APPENDIX B—SCATTER ANALYSIS RESULTS FOR FEDERAL AVIATION
ADMINISTRATION LOAD-ENHANCEMENT FACTOR DATABASE
This appendix contains the static data scatter analysis results of Federal Aviation Administration
lamina variability method (FAA-LVM) database. This analysis includes data for T700/#2510
unidirectional tape (T700-UT), AS4C/MTM45 unidirectional tape (AS4C-UT), AS4C/MTM45
5-harness satin-weave fabric (AS4C-5HS), T700/E765 unidirectional tape (E765-UT), and
T300/E765 plain-weave fabric material systems. The scatter analysis of T700/#2510 plainweave fabric (T700-PW) data belonging to the FAA-LVM database is included in section 4.1.2
of the main document.
B.1 STATIC SCATTER ANALYSIS OF T700-UT.
This section contains the shape parameters of 853 T700-UT specimens from 47 data sets
obtained from the FAA-LVM database. Tables B-1, B-2, and B-3 contain shape parameters
corresponding to static-strength distributions of individual test methods and environmental
conditions of hard (50/40/10), quasi-isotropic (25/50/25), and soft (10/80/10) laminates,
respectively. The scatter analysis in section 4.1.4 of the main document was conducted by
analyzing the shape parameters in these three tables.
Table B-1. Weibull Parameters for Static-Strength Distributions of
50/40/10 T700-UT (FAA-LVM)
Test Description
Single-shear bearing tension
Double-shear bearing tension
Bearing-bypass 50% compression
Bearing-bypass 50% tension [t/D = 0.320]
Bearing-bypass 50% tension [t/D = 0.384]
Bearing-bypass 50% tension [t/D = 0.640]
Bearing-bypass 50% tension [t/D = 0.480]
Unnotched tension
Unnotched compression
Open-hole compression
Filled-hole tension
Filled-hole tension
Filled-hole tension
V-notched rail shear
Open-hole tension [w/D = 3]
Open-hole tension [w/D = 4]
Open-hole tension [w/D = 8]
CTD = Cold temperature dry
RTA = Room temperature ambient
Test
Environment
RTA
RTA
RTA
RTA
RTA
RTA
RTA
RTA
RTA
RTA
CTD
RTA
ETW
RTA
RTA
RTA
RTA
Shape Parameter,
ETW = Elevated temperature wet
t/D = Thickness to diameter
B-1
α̂
52.085
40.405
38.043
41.398
44.098
42.257
59.840
20.702
33.121
22.705
36.224
17.585
20.516
22.705
22.553
20.420
32.794
Number of
Specimens
19
20
15
18
18
17
15
19
19
21
18
18
19
18
18
18
19
w/D = Width to diameter
Table B-2. Weibull Parameters for Static-Strength Distributions of
25/50/25 T700-UT (FAA-LVM)
Test Description
Double-shear bearing tension
Double-shear bearing tension
Double-shear bearing tension
Single-shear bearing tension
Single-shear bearing tension
Single-shear bearing tension
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Unnotched compression
Open-hole compression
Open-hole compression
Open-hole compression
V-notched rail shear
Test Environment
CTD
RTA
ETW
CTD
RTA
ETW
RTA
RTA
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
RTA
Shape Parameter, α̂
31.499
10.310
41.221
37.563
14.452
18.924
74.669
57.956
37.058
25.000
27.714
45.075
39.641
32.490
23.026
22.034
34.869
37.455
27.930
22.122
9.054
Number of Specimens
22
19
18
18
18
18
15
15
18
18
21
18
18
18
18
18
18
21
18
22
18
Table B-3. Weibull Parameters for Static-Strength Distributions of
10/80/10 T700-UT (FAA-LVM)
Test Description
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched compression
Open-hole compression
V-notched rail shear
V-notched rail shear
V-notched rail shear
Test Environment
RTA
RTA
RTA
RTA
RTA
RTA
CTD
RTA
ETW
B-2
Shape Parameter, α̂
64.772
49.444
67.241
20.186
26.157
41.378
7.262
19.659
10.595
Number of
Specimens
15
15
18
18
18
18
18
19
18
B.2 STATIC SCATTER ANALYSIS OF AS4C/MTM45 UNIDIRECTIONAL TAPE.
This section contains the shape parameters of 1151 AS4C-UT specimens from 86 data sets
obtained from the FAA-LVM database. Tables B-4 through B-6 contain shape parameters
corresponding to static-strength distributions of individual test methods and environmental
conditions of hard (50/40/10), quasi-isotropic (25/50/25), and soft (10/80/10) laminates,
respectively. In addition, shape parameters corresponding to AS4C-UT lamina data are included
in table B-7. The scatter analysis in section 4.1.5 of the main document was conducted by
analyzing the shape parameters in these three tables.
Table B-4. Weibull Parameters for Static-Strength Distributions of
50/40/10 AS4C-UT (FAA-LVM)
Test Description
Unnotched tension
Unnotched tension
Unnotched tension
Open-hole tension
Open-hole tension
Open-hole tension
Filled-hole tension
Filled-hole tension
Unnotched compression
Unnotched compression
Open-hole compression
Open-hole compression
Filled-hole compression
Filled-hole compression
Double-shear bearing
Double-shear bearing
Test Environment
CTD
RTD
ETW2
CTD
RTD
ETW2
CTD
RTD
RTD
ETW
RTD
ETW2
RTD
ETW2
RTD
ETW2
CTD = Cold temperature dry
RTD = Room temperature dry
ETW2 = Elevated temperature 220°F, wet
B-3
Shape Parameter, α̂
36.6884
39.6101
105.7766
8.2884
46.3690
60.1876
18.7167
18.1374
46.4812
39.6044
51.4916
28.7315
7.6270
13.1700
52.4169
30.4239
Number of
Specimens
7
6
6
18
6
6
6
6
6
7
6
19
7
21
6
20
Table B-5. Weibull Parameters for Static-Strength Distributions of
25/50/25 AS4C-UT (FAA-LVM)
Test Description
Test Environment Shape Parameter, α̂
Unnotched tension
CTD
36.9107
Unnotched tension
RTD
36.7091
Unnotched tension
ETW2
73.1509
Open-hole tension
CTD
27.1127
Open-hole tension
RTD
35.9034
Open-hole tension
ETW
45.0771
Open-hole tension
ETW2
29.5619
Filled-hole tension
CTD
30.4239
Filled-hole tension
RTD
32.5909
Unnotched compression
RTD
35.8363
Unnotched compression
ETW
45.5883
Unnotched compression
ETW2
25.0000
Open-hole compression
RTD
24.9999
Open-hole compression
ETW
23.6168
Open-hole compression
ETW2
28.0623
Filled-hole compression
RTD
24.9999
Filled-hole compression
ETW2
21.4617
Double-shear bearing
RTD
21.0857
Double-shear bearing
ETW2
22.9000
Compression after impact
RTD
16.0114
Number of
Specimens
18
18
6
15
16
5
18
18
6
19
7
18
14
6
17
6
7
18
18
6
CTD = Cold temperature dry
RTD = Room temperature dry
ETW2 = Elevated temperature 220°F, wet
Table B-6. Weibull Parameters for Static-Strength Distributions of
10/80/10 AS4C-UT (FAA-LVM)
Test Description
Unnotched tension
Unnotched tension
Unnotched tension
Open-hole tension
Open-hole tension
Test Environment
CTD
RTD
ETW2
CTD
RTD
B-4
Shape Parameter, α̂
35.1583
31.4154
47.2541
48.8726
47.2759
Number of
Specimens
6
6
6
17
5
Table B-6. Weibull Parameters for Static-Strength Distributions of
10/80/10 AS4C-UT (FAA-LVM) (Continued)
Test Description
Open-hole tension
Filled-hole tension
Filled-hole tension
Filled-hole tension
Unnotched compression
Unnotched compression
Open-hole compression
Open-hole compression
Filled-hole compression
Filled-hole compression
Double-shear bearing
Double-shear bearing
Short-beam shear
Short-beam shear
Short-beam shear
Test Environment
ETW2
CTD
RTD
ETW2
RTD
ETW2
RTD
ETW2
RTD
ETW2
RTD
ETW2
RTD
ETW
ETW2
Shape Parameter, α̂
38.5491
60.3965
40.4534
38.2105
33.1313
47.7841
33.2611
27.2031
38.5239
27.6622
31.2818
18.3542
22.0431
23.9989
27.0304
Number of
Specimens
6
6
6
7
6
6
5
18
5
21
6
12
18
6
18
CTD = Cold temperature dry
RTD = Room temperature dry
ETW2 = Elevated temperature 220°F, wet
Table B-7. Weibull Parameters for Static-Strength Distributions of
AS4C-UT Lamina Data (FAA-LVM)
Lay-Up
100/0/0
0/0/100
Test Description
Longitudinal tension
Longitudinal tension
Longitudinal tension
Longitudinal tension
Transverse tension
Transverse tension
Transverse tension
Transverse tension
Test
Environment
CTD
RTD
ETW
ETW2
CTD
RTD
ETW
ETW2
B-5
Shape Parameter,
α̂
13.8596
15.0419
16.9554
18.9761
13.2656
13.7326
12.5037
16.5834
Number of
Specimens
20
18
12
12
20
20
23
21
Table B-7. Weibull Parameters for Static-Strength Distributions of
AS4C-UT Lamina Data (FAA-LVM) (Continued)
Lay-Up
0/0/100
0/100/0
100/0/0
50/0/50
50/0/50
Test Description
Transverse compression
Transverse compression
Transverse compression
Transverse compression
In-plane shear
In-plane shear
In-plane shear
In-plane shear
Short-beam shear
Short-beam shear
Short-beam shear
Short-beam shear
Short-beam shear
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Unnotched compression
Unnotched compression
Unnotched compression
Test
Environment
CTD
RTD
ETW
ETW2
CTD
RTD
ETW
ETW2
CTD
RTD
ETD
ETW
ETW2
CTD
RTD
ETW
ETW2
CTD
RTD
ETD
ETW
ETW2
Shape Parameter,
α̂
22.5142
32.6974
31.1187
23.3048
30.8707
43.3159
37.3768
31.4522
30.0698
32.5469
26.2652
36.3365
19.1586
18.9930
19.3424
16.0501
17.8260
14.1693
28.1360
20.7004
20.0349
12.5308
Number of
Specimens
19
18
18
19
18
18
17
19
20
20
20
21
18
18
18
18
17
22
18
18
18
18
CTD = Cold temperature dry
RTD = Room temperature dry
ETW2 = Elevated temperature 220°F, wet
B.3 STATIC SCATTER ANALYSIS OF AS4C/MTM45 5-HARNESS SATIN-WEAVE
FABRIC.
This section contains the shape parameters of 1083 AS4C-5HS specimens from 78 data sets
obtained from the FAA-LVM database. Tables B-8 through B-10 contain shape parameters
corresponding to static-strength distributions of individual test methods and environmental
conditions of hard (40/20/40), quasi-isotropic (25/50/25), and soft (10/80/10) laminates,
respectively. In addition, shape parameters corresponding to AS4C-5HS lamina data are
included in table B-11. The scatter analysis in section 4.1.6 of the main document was
conducted by analyzing the shape parameters in these three tables.
B-6
Table B-8. Weibull Parameters for Static-Strength Distributions of
40/20/40 AS4C-5HS (FAA-LVM)
Test Description
Test Environment
Unnotched tension
CTD
Unnotched compression
RTD
Unnotched compression
ETW2
Open-hole tension
CTD
Open-hole tension
RTD
Open-hole tension
ETW2
Filled-hole tension
CTD
Filled-hole tension
RTD
Open-hole compression
RTD
Open-hole compression
ETW2
Filled-hole compression
RTD
Filled-hole compression
ETW2
Double-shear bearing
RTD
Double-shear bearing
ETW2
Number of
Specimens
6
6
6
21
7
6
6
6
6
18
6
20
6
24
Shape Parameter, α̂
50.3322
13.4587
23.7333
25.5530
33.2549
28.5810
20.9094
29.0854
32.2012
27.0483
14.1911
19.3906
34.7686
15.5135
CTD = Cold temperature dry
RTD = Room temperature dry
ETW2 = Elevated temperature 220°F, wet
Table B-9. Weibull Parameters for Static-Strength Distributions of
25/50/25 AS4C-5HS (FAA-LVM)
Test Description
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Unnotched compression
Open-hole tension
Open-hole tension
Open-hole tension
Open-hole tension
Filled-hole tension
Filled-hole tension
Test Environment
CTD
RTD
ETW2
RTD
ETW
ETW2
RTD
CTD
ETW
ETW2
RTD
CTD
B-7
Shape Parameter, α̂
62.6999
34.3319
44.9484
20.4637
19.1119
19.6297
40.4981
24.3826
86.9983
28.7676
41.9103
34.6634
Number of
Specimens
23
18
6
18
6
18
18
18
6
18
6
18
Table B-9. Weibull Parameters for Static-Strength Distributions of
25/50/25 AS4C-5HS (FAA-LVM) (Continued)
Test Description
Open-hole compression
Open-hole compression
Open-hole compression
Filled-hole compression
Filled-hole compression
Double-shear bearing
Double-shear bearing
Short-beam shear
Short-beam shear
Short-beam shear
Test Environment
RTD
ETW
ETW2
RTD
ETW2
RTD
ETW2
RTD
ETW
ETW2
Shape Parameter, α̂
29.5250
15.5124
24.5294
27.7953
31.8401
13.9867
13.6569
35.7819
61.9031
57.5735
CTD = Cold temperature dry RTD = Room temperature dry
Number of
Specimens
18
6
18
6
18
18
23
18
6
18
ETW2 = Elevated temperature 220°F, wet
Table B-10. Weibull Parameters for Static-Strength Distributions of
10/80/10 AS4C-5HS (FAA-LVM)
Test Description
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Open-hole tension
Open-hole tension
Open-hole tension
Filled-hole tension
Filled-hole tension
Filled-hole tension
Open-hole compression
Open-hole compression
Filled-hole compression
Filled-hole compression
Double-shear bearing
Double-shear bearing
CTD = Cold temperature dry
Test Environment
CTD
RTD
ETW2
RTD
ETW2
CTD
RTD
ETW2
RTD
CTD
ETW2
RTD
ETW2
RTD
ETW2
RTD
ETW2
Shape Parameter, α̂
36.4157
74.8199
26.8754
23.2597
32.6546
41.9517
53.7592
40.8813
53.8676
64.3808
43.5941
32.1458
40.5532
64.9168
41.6289
21.9478
14.6494
RTD = Room temperature dry
B-8
Number of
Specimens
6
6
6
6
6
18
9
6
6
6
6
6
18
6
18
6
24
ETW2 = Elevated temperature 220°F, wet
Table B-11. Weibull Parameters for Static-Strength Distributions of
50/0/50 AS4C-5HS (FAA-LVM)
Test Description
Warp tension
Warp tension
Warp tension
Warp tension
Filled tension
Filled tension
Filled tension
Filled tension
Warp compression
Warp compression
Warp compression
Warp compression
Filled compression
Filled compression
Filled compression
Filled compression
Filled compression
In-plane Shear
In-plane Shear
In-plane Shear
In-plane Shear
Short-beam shear
Short-beam shear
Short-beam shear
Short-beam shear
Test Environment
CTD
RTA
ETW
ETW2
CTD
RTA
ETW
ETW2
CTD
RTA
ETW
ETW2
CTD
RTA
ETD
ETW
ETW2
CTD
RTA
ETW
ETW2
CTD
RTA
ETW
ETW2
RTA = Room temperature ambient
RTD = Room temperature dry
Shape Parameter, α̂
41.7447
31.7590
34.1953
25.6064
27.5161
31.6162
36.2862
25.4495
11.9994
16.1442
12.2082
9.7697
17.3767
18.1475
13.3973
15.3266
18.4131
14.1296
14.7233
61.6350
13.9587
24.7163
36.7471
26.2523
28.0774
Number of
Specimens
19
22
20
22
18
18
21
19
18
19
18
18
18
18
18
18
18
16
16
16
16
19
17
18
18
CTD = Cold temperature dry
ETW2 = Elevated temperature 220°F, wet
B.4 STATIC SCATTER ANALYSIS OF T700/E765 UNIDIRECTIONAL TAPE.
This section contains the shape parameters of 834 E765-UT specimens from 47 data sets
obtained from the FAA-LVM database. Tables B-12 through B-14 contain shape parameters
corresponding to static-strength distributions of individual test methods and environmental
conditions of hard (50/40/10), quasi-isotropic (25/50/25), and soft (10/80/10) laminates,
respectively. The scatter analysis in section 4.1.7 of the main document was conducted by
analyzing the shape parameters in these three tables.
B-9
Table B-12. Weibull Parameters for Static-Strength Distributions of
50/40/10 E765-UT (FAA-LVM)
Test Description
Single-shear bearing tension
Double-shear bearing tension
Bearing-bypass 50% compression
Bearing-bypass 50% tension
[t/D = 0.320]
Bearing-bypass 50% tension
[t/D = 0.384]
Bearing-bypass 50% tension
[t/D = 0.640]
Bearing-bypass 50% tension
[t/D = 0.480]
Unnotched tension
Unnotched compression
Open-hole compression
Filled-hole tension
Filled-hole tension
Filled-hole tension
V-notched rail shear
Open-hole tension [w/D = 3]
Open-hole tension [w/D = 4]
Open-hole tension [w/D = 8]
Test Environment
RTA
RTA
RTA
RTA
Shape Parameter, α̂
30.814
17.957
28.550
30.861
Number of
Specimens
18
21
13
17
RTA
46.294
18
RTA
24.909
18
RTA
26.362
15
RTA
RTA
RTA
CTD
RTA
ETW
RTA
RTA
RTA
RTA
7.347
8.897
21.189
31.088
33.685
18.421
19.226
35.484
77.305
27.486
20
18
20
18
18
18
18
18
16
18
RTA = Room temperature ambient
CTD = Cold temperature dry
ETW = Elevated temperature wet
Table B-13. Weibull Parameters for Static-Strength Distributions of
25/50/25 E765-UT (FAA-LVM)
Test Description
Double-shear bearing tension
Double-shear bearing tension
Double-shear bearing tension
Single-shear bearing tension
Test Environment
CTD
RTA
ETW
CTD
B-10
Shape Parameter, α̂
13.077
28.116
8.356
17.509
Number of
Specimens
19
18
21
18
Table B-13. Weibull Parameters for Static-Strength Distributions of
25/50/25 E765-UT (FAA-LVM) (Continued)
Test Description
Single-shear bearing tension
Single-shear bearing tension
Bearing-bypass 50% tension
Bearing-bypass 50%
compression
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Unnotched compression
Open-hole compression
Open-hole compression
Open-hole compression
V-notched rail shear
RTA = Room temperature ambient
Test Environment
RTA
ETW
RTA
RTA
Shape Parameter, α̂
26.754
31.724
40.112
30.226
Number of
Specimens
18
20
16
15
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
RTA
37.529
50.949
48.482
14.611
25.580
24.874
18.276
28.061
10.442
18.715
50.933
13.645
10.840
18
18
18
19
14
18
18
17
18
19
18
20
18
CTD = Cold temperature dry
ETW = Elevated temperature wet
Table B-14. Weibull Parameters for Static-Strength Distributions of
10/80/10 E765-UT (FAA-LVM)
Test Description
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched compression
Open hole compression
V-notched rail shear
V-notched rail shear
V-notched rail shear
RTA = Room temperature ambient
Test Environment
RTA
RTA
RTA
RTA
RTA
RTA
CTD
RTA
ETW
Shape Parameter, α̂
38.362
39.852
32.358
40.424
27.476
28.881
6.335
16.180
11.573
CTD = Cold temperature dry
B-11
Number of
Specimens
15
14
18
15
18
20
18
18
18
ETW = Elevated temperature wet
B.5 STATIC SCATTER ANALYSIS OF T300/E765 3K PLAIN-WEAVE FABRIC.
This section contains the shape parameters of 722 E765-PW specimens from 48 data sets
obtained from the FAA-LVM database. Tables B-15 through B-17 contain shape parameters
corresponding to static-strength distributions of individual test methods and environmental
conditions of hard (40/20/40), quasi-isotropic (25/50/25), and soft (10/80/10) laminates,
respectively. The scatter analysis in section 4.1.8 of the main document was conducted by
analyzing the shape parameters in these three tables.
Table B-15. Weibull Parameters for Static-Strength Distributions of
(40/20/40 E765-PW (FAA-LVM)
Test Description
Single-shear bearing tension
Double-shear bearing tension
Bearing-bypass 50% compression
Bearing-bypass 50% tension
[t/D = 0.320]
Bearing-bypass 50% tension
[t/D = 0.384]
Bearing-bypass 50% tension
[t/D = 0.640]
Bearing-bypass 50% tension
[t/D = 0.480]
Unnotched tension
Unnotched compression
Open-hole compression
Filled-hole tension
Filled-hole tension
Filled-hole tension
V-notched rail shear
Open-hole tension [w/D = 3]
Open-hole tension [w/D = 4]
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 8]
Test Environment
RTA
RTA
RTA
RTA
Shape Parameter, α̂
14.391
39.826
34.680
49.915
Number of
Specimens
15
13
9
12
RTA
43.274
12
RTA
2.986
6
RTA
27.165
12
RTA
RTA
RTA
CTD
RTA
ETW
RTA
RTA
RTA
RTA
RTA
27.277
23.693
31.510
20.391
45.561
35.757
65.531
34.781
51.052
29.581
32.084
13
13
12
12
12
12
12
12
13
12
12
RTA = Room temperature ambient
CTD = Cold temperature dry
ETW = Elevated temperature wet
B-12
Table B-16. Weibull Parameters for Static-Strength distributions of
25/50/25 E765-PW (FAA-LVM)
Test Description
Double-shear bearing tension
Double-shear bearing tension
Double-shear bearing tension
Single-shear bearing tension
Single-shear bearing tension
Single-shear bearing tension
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Open-hole tension [w/D = 6]
Unnotched tension
Unnotched tension
Unnotched tension
Unnotched compression
Unnotched compression
Unnotched compression
Open-hole compression
Open-hole compression
Open-hole compression
V-notched rail shear
Test Environment
CTD
RTA
ETW
CTD
RTA
ETW
RTA
RTA
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
CTD
RTA
ETW
RTA
Shape Parameter, α̂
19.297
38.719
28.345
12.403
22.420
19.406
40.069
34.458
19.337
18.427
23.693
43.886
43.824
42.304
26.180
33.422
20.017
19.446
25.503
12.103
20.670
Number of
Specimens
20
20
19
18
18
18
10
8
18
18
18
18
18
21
18
15
18
19
18
19
18
RTA = Room temperature ambient CTD = Cold temperature dry ETW = Elevated temperature wet
Table B-17. Weibull Parameters for Static-Strength Distributions of
(10/80/10 E765-PW (FAA-LVM)
Test Description
Bearing-bypass 50% tension
Bearing-bypass 50% compression
Open-Hole tension [w/D = 6]
Unnotched tension
Unnotched compression
Open-hole compression
V-notched rail shear
V-notched rail shear
V-notched rail shear
RTA = Room temperature ambient
Test Environment
RTA
RTA
RTA
RTA
RTA
RTA
CTD
RTA
ETW
Shape Parameter, α̂
8.873
25.206
28.001
44.073
41.200
39.190
12.622
7.238
14.961
Number of
Specimens
10
13
18
18
18
18
12
19
15
CTD = Cold temperature dry ETW = Elevated temperature wet
B-13/B-14
APPENDIX C—SCATTER ANALYSIS FOR LIFE AND
LOAD-ENHANCEMENT FACTORS
Figure C-1 illustrates the procedure for analyzing stress to number of cycle (S/N) data using
individual Weibull, joint Weibull, and Sendeckyj analysis for generating the fatigue-life shape
parameter for a particular test configuration ( α̂ IW , α̂ JW , and α̂ Sendeckyj , respectively). The details
for each method are discussed in section 3.1 of the main document. As shown in this figure, the
static data, if applicable, are analyzed using individual Weibull for generating the static-strength
shape parameter ( α̂ SS ) for the same test configuration.
ThV-notched rail sheare procedure for generating the static-strength (  R ) and fatigue-life (  L )
shape parameters for calculating the life factor and load-enhancement factors for a particular
structure is illustrated in figure C-2. This process requires analysis of static/residual strength
data and fatigue (S/N) data for multiple test configurations that represent critical design details of
the structure. The fatigue-life shape parameter ( α̂ FL ) for each design detail or test configuration
can be obtained using one of the three methods ( α̂ IW , α̂ JW , or α̂ Sendeckyj ) shown in figure C-1. The
details for calculating life factor, load factor, and load-enhancement factors are included in
sections 3.2 through 3.4 of the main document.
C-1
Figure C-1. Process for Obtaining the Fatigue-Life Shape Parameters for an Individual
Test Configuration
C-2
1
2
SS
i
SS
SS
Static Data
Data Set i+1
i+1
SS
Data Set i+2
i+2
SS
Weibull Analysis
j
Data Set j
SS
SN Data – Set 1
SN Data – Set 2
SN Data – Set i
Static Strength
Static Strength
Static Strength
Fatigue Data
Fatigue Data
Fatigue Data
Stress Level 1
Stress Level 1
Stress Level 1
Stress Level 2
Stress Level 2
Stress Level 2
Stress Level m
Stress Level m
Stress Level m
Weibull
Analysis
R
Individual Weibull Analysis
Joint Weibull Analysis
Sendeckyj Analysis
1
FL
2
FL
i
FL
Weibull
Analysis
L
Figure C-2. Process for Generating the Static-Strength and Fatigue-Life Shape Parameters for
Calculating the Life Factor and Load-Enhancement Factors for a Particular Structure
C-3/C-4
Download